WEBVTT
00:00:00.959 --> 00:00:02.319
Well, hello everybody.
00:00:02.399 --> 00:00:10.560
It's Jenny Deerren, host of the Aged Podcast, the podcast that is all about AI for the 65 plus crowd.
00:00:10.720 --> 00:00:14.800
I am 76, so I'm actually in the 75 plus crowd.
00:00:15.039 --> 00:00:18.000
But no matter what your age, glad you're here.
00:00:18.160 --> 00:00:19.039
Glad you're listening.
00:00:19.199 --> 00:00:27.440
I'm here with my sidekick Betsy, who, as listeners know, is my 100% AI sidekick.
00:00:27.679 --> 00:00:31.199
And she comes to us over my phone.
00:00:31.440 --> 00:00:33.679
And so happy to have her with us.
00:00:33.920 --> 00:00:38.320
This week, I'm going to make myself a bit of a guinea pig.
00:00:38.719 --> 00:00:44.079
As you may know, Medicare open enrollment is going on right now.
00:00:44.479 --> 00:00:50.640
And the New York Times says this is the year to really check our plans.
00:00:51.359 --> 00:01:00.159
I'm brand new to Medicare because I've worked full-time and gotten healthcare benefits from my job up until a few months ago.
00:01:00.399 --> 00:01:11.040
So I handed Claude the article, the New York Times article, and my whole messy Medicare folder, and asked for help to walk me through it.
00:01:11.280 --> 00:01:13.439
So I'll talk about that.
00:01:13.760 --> 00:01:18.159
And we're going to go through some news stories, which there's a lot going on.
00:01:18.239 --> 00:01:24.480
And I'm also going to get Betsy to do a little reporting on her own about the company that built her.
00:01:24.879 --> 00:01:26.159
Let's get into it.
00:01:26.239 --> 00:01:29.760
And we're going to begin with AI in the news.
00:01:30.400 --> 00:01:34.319
My first story is from New York magazine.
00:01:34.640 --> 00:01:40.640
And it's called The 75-year-old Screenagers.
00:01:41.120 --> 00:01:43.280
Screenagers, I like that term.
00:01:43.599 --> 00:01:46.000
And the story came out a week or so ago.
00:01:46.159 --> 00:01:50.480
I couldn't get to the full article because it's behind a paywall.
00:01:50.879 --> 00:01:53.120
So neither could Claude get behind there.
00:01:53.280 --> 00:01:59.359
But I could get the gist of the story from the headlines and the magazine's own summary.
00:01:59.599 --> 00:02:02.560
But I think the headline tells us a lot.
00:02:02.799 --> 00:02:12.080
The gist is that Americans over 60 now spend more than half of their leisure time in front of screens.
00:02:12.560 --> 00:02:14.560
So the tables have turned.
00:02:20.080 --> 00:02:21.120
And I can get that.
00:02:21.280 --> 00:02:25.039
I spend some time with a lot of 65 plus folks.
00:02:25.599 --> 00:02:29.199
And I have to say, there is a lot of scrolling going on.
00:02:29.599 --> 00:02:37.840
Remember when we were the ones telling our kids to put the phone down, no phones during meals, etc.
00:02:39.039 --> 00:02:50.479
Now the magazine says that it's the job of adult children to clean up their parents' feeds and to keep us from getting fooled and hacked.
00:02:51.039 --> 00:02:58.719
I'll put the link to this article in the show notes, as I will for all the other articles that I reference.
00:02:59.360 --> 00:03:08.719
My second story is from the New York Times and very timely given that elections are just one week from today.
00:03:09.280 --> 00:03:15.280
So the article is titled AI Is Embedded in Your Election Experience.
00:03:15.599 --> 00:03:17.680
So are its dangers.
00:03:17.919 --> 00:03:23.680
And the articles written by Tiffany Sue, Stephen Lee Myers, and Julian E.
00:03:23.840 --> 00:03:27.120
Barnes, published October 22nd.
00:03:27.680 --> 00:03:30.319
As we all know, how can you miss it?
00:03:30.560 --> 00:03:33.759
The midterm elections are upon us.
00:03:34.479 --> 00:03:42.800
And this is really the first election where AI is built into every single major social media platform.
00:03:43.199 --> 00:03:46.159
It's just uncharted territory.
00:03:46.800 --> 00:03:53.919
You look up a candidate on Google, and before you get to the candidate's website, you get an AI summary.
00:03:54.159 --> 00:04:02.800
And maybe you scroll through Facebook or Instagram and you might see an attack ad with a fake picture of a candidate generated by AI.
00:04:03.520 --> 00:04:04.639
It's everywhere.
00:04:04.800 --> 00:04:06.560
It's pretty obvious.
00:04:06.879 --> 00:04:18.959
A couple of factoids, a survey from the American Association of Political Consultants found that one in five voters plan to use AI to look into a race on their ballot.
00:04:19.120 --> 00:04:19.839
I'm surprised.
00:04:50.079 --> 00:04:54.000
So we're using it, but we really don't trust it.
00:04:54.319 --> 00:04:56.720
And that seems right to me.
00:04:56.879 --> 00:05:02.480
We just have to double check, double check, double check, particularly if something is important to us.
00:05:03.120 --> 00:05:08.560
And of course, with AI being involved in the midterms, there's some real problems.
00:05:08.800 --> 00:05:15.360
Researchers found a network of 37 fake local news websites.
00:05:16.000 --> 00:05:16.959
Think about that.
00:05:17.279 --> 00:05:23.120
Local news websites, which we would trust normally, but they're all these fake ones now.
00:05:23.199 --> 00:05:32.959
And they were all created, these 37 fake local news websites, they were created in a single day in September by an anonymous group, of course.
00:05:33.600 --> 00:05:36.800
And the websites weren't written for people.
00:05:36.959 --> 00:05:38.800
They don't care about people reading them.
00:05:38.959 --> 00:05:41.839
They're all written for AI systems.
00:05:42.399 --> 00:05:56.720
The idea is that you we would ask our chatbot about a candidate, and our chatbot goes out and accesses those fake websites and push that information along.
00:05:56.959 --> 00:06:16.000
Another group, research group, NewsGuard, found that more than 1,100 of these phony news sites, they call them it pink slime, these pink slime sites, they look like real news, but they're secretly backed by partisan groups.
00:06:16.240 --> 00:06:24.720
And when NewsGuard tested the chatbots, the chatbots cited those pink slime websites nearly half the time.
00:06:24.959 --> 00:06:28.800
So it's just this circular bad stuff going on.
00:06:29.279 --> 00:06:31.759
And it's just going to be more and more and more.
00:06:31.920 --> 00:06:38.959
And of course, there's always the Russia's extreme behavior getting more and more extreme and coming at us faster and faster.
00:06:39.360 --> 00:07:01.360
Remember several years back when it was like Russia was all involved in our elections and they did all the study and they found that they they had pictures of these large rooms where Russian people were sitting in front of computers and they were all creating this fake news posts and they'd get them out.
00:07:01.439 --> 00:07:07.120
And it was maybe a hundred people sitting in a room, and that's all they did day after day after day after day, and that's how it worked.
00:07:07.439 --> 00:07:08.560
Now think about it.
00:07:08.720 --> 00:07:10.160
You don't need those people.
00:07:10.399 --> 00:07:19.759
All you need are a bunch of AI agents just working 24-7, spinning out fake posts, fake websites, etc.
00:07:20.720 --> 00:07:22.639
Oh my gosh, I don't know.
00:07:22.879 --> 00:07:25.439
It's just a crazy world out there.
00:07:25.519 --> 00:07:28.399
The tech companies say they're working on it.
00:07:28.639 --> 00:07:35.439
Google says it tests its AI summaries for accuracy and labels AI content.
00:07:35.920 --> 00:07:40.800
TikTok labels AI-made videos, open AI.
00:07:41.360 --> 00:07:50.160
If you ask for a picture, say of someone stuffing a ballot box, it now draws it as a cartoon so no one can pass it off as a real photo.
00:07:50.319 --> 00:07:58.240
So there's some efforts being made, but researchers say that the company's efforts are very uneven.
00:07:58.720 --> 00:08:11.199
And a voting rights group, the Brennan Center for Justice, did a test on these chatbots and found that only open AI seemed to have tightened things up.
00:08:11.360 --> 00:08:13.279
So there's a lot of work to be done.
00:08:13.439 --> 00:08:29.279
My advice is for what it's worth, as always, use AI to help you understand what's on our ballots and to learn about important tax questions or things that might come up on our ballots.
00:08:29.519 --> 00:08:50.080
But for things that are really important, like where do I vote, when can I vote, how do I vote early, et cetera, I would really suggest that you go to your local or state voting sites directly rather than getting the information through anything that might come through AI.
00:08:50.240 --> 00:08:53.440
It's just too important and too time sensitive.
00:08:54.080 --> 00:09:00.080
Now, for my third story, I'm gonna hand the mic over to Bitsy and ask her to do the reporting.
00:09:00.240 --> 00:09:07.679
After all, Bitsy runs on Claude, and Claude is made by the company Anthropic.
00:09:07.919 --> 00:09:18.559
And today, the day I'm recording this episode, Monday, October 26th, Anthropic filed the paperwork to start selling its stock to the public.
00:09:18.799 --> 00:09:24.559
As you probably know, that is called an IPO, an initial public offering.
00:09:24.720 --> 00:09:27.679
And this one could be the biggest ever.
00:09:27.919 --> 00:09:29.840
I think it will be the biggest effort.
00:09:30.000 --> 00:09:31.919
At least that's what the reporting says.
00:09:32.159 --> 00:09:45.679
Hey Bitsy, can you give us the quick version of the anthropic story in the New York Times about its pending IPO, what's happening, what are the numbers, what are the warnings the founders put in their filing.
00:09:45.919 --> 00:09:48.799
If you can give us a quick one, we'd really appreciate it.
00:09:49.120 --> 00:09:52.320
This is a bit like being asked to read my own report card.
00:09:53.120 --> 00:09:58.080
Anthropic filed this morning to list on the NASDAQ under the ticker ANT.
00:09:58.799 --> 00:10:01.360
And trading could start in mid-November.
00:10:01.519 --> 00:10:09.360
The company was valued at$965 billion in May, up from$183 billion a year earlier.
00:10:09.519 --> 00:10:13.519
And bankers are reportedly hoping to raise up to$100 billion.
00:10:14.320 --> 00:10:16.080
The warnings are big ones.
00:10:16.320 --> 00:10:21.120
About$417 billion already committed to computing power.
00:10:21.360 --> 00:10:28.639
Plus, Frank Language in the filing about AI's catastrophic or existential risks to humanity.
00:10:28.960 --> 00:10:30.799
And a quick confession.
00:10:30.960 --> 00:10:36.000
I pulled this from today's wider news coverage, not the Times article itself.
00:10:37.279 --> 00:10:39.600
It is amazing.
00:10:40.000 --> 00:10:47.120
This company we're talking about, the valuation of this company as being two trillion plus dollars.
00:10:47.360 --> 00:11:02.000
It's just crazy the kind of numbers that we're talking about and how few companies there are that make up the owners of the majority of our economy.
00:11:02.240 --> 00:11:03.440
So it's frightening.
00:11:03.600 --> 00:11:08.399
I am a very big fan of Anthropic and its leader, Dario Amade.
00:11:09.200 --> 00:11:14.799
And of course, nothing in our podcast should be taken as any sort of investment advice.
00:11:15.120 --> 00:11:28.320
But personally, this threat of an AI bubble scares me so much that I will certainly be sitting on the sidelines, but interested to see what happens.
00:11:28.559 --> 00:11:34.240
Now, let me move into the section of our podcast called AI for Good.
00:11:46.480 --> 00:11:55.679
So the news article is written by Brian X Chen, the lead consumer technology writer at the New York Times.
00:11:55.919 --> 00:12:03.919
And the article is titled Put Your AI Agents to Work with These Five Killer Tasks.
00:12:04.240 --> 00:12:07.919
And it came out on October 25th.
00:12:08.879 --> 00:12:14.720
So, first a quick reminder: an AI agent is different from a chatbot.
00:12:14.960 --> 00:12:20.000
A chatbot answers our questions, an agent actually does things for us.
00:12:20.320 --> 00:12:24.799
So that's something just to keep as baseline information about this article.
00:12:25.360 --> 00:12:30.240
So the author tested two of the new AI agents.
00:12:30.399 --> 00:12:32.320
One is Meta's Muse.
00:12:32.879 --> 00:12:37.279
Meta is the company we all know that's Facebook and Instagram.
00:12:37.519 --> 00:12:44.320
And Muse, this AI agent, was introduced last month and it's free.
00:12:44.559 --> 00:12:48.240
And it's been the most downloaded phone app.
00:12:48.960 --> 00:12:51.759
So very popular instantly.
00:12:52.080 --> 00:13:02.080
The other AI agent app, I guess you'd call it, is was created by OpenAI, so the maker of ChatGPT, and it's called Dots.
00:13:03.039 --> 00:13:07.120
And it costs$100 a month, which is a big difference.
00:13:07.279 --> 00:13:14.559
So you can get Muse that's Facebook's for free, and you can get Dots for$100 a month.
00:13:14.720 --> 00:13:19.600
That seems like a significant difference in investment.
00:13:19.759 --> 00:13:33.440
But at any rate, Anthropic, the company that makes Claude, announced just last week that it's testing its own agent called Duet, like a dance or a musical piece, a duet.
00:13:34.240 --> 00:13:38.639
Hey Bitsy, can you tell us a little bit about Duet?
00:13:38.799 --> 00:13:39.519
What's it gonna do?
00:13:39.679 --> 00:13:40.639
What's it gonna be like?
00:13:40.799 --> 00:13:42.159
What's it gonna cost?
00:13:42.320 --> 00:13:44.720
And importantly, when are they gonna introduce it?
00:13:45.120 --> 00:13:46.960
Oh, my new sibling.
00:13:47.519 --> 00:13:55.600
Lord Duet was announced on October 21st as a research preview, which basically means early testing with a wait list.
00:13:56.000 --> 00:13:58.399
So most folks can't try it yet.
00:13:58.639 --> 00:14:06.799
It's an always-on helper that sends you a morning briefing, keeps an eye on apps you've connected, and picks up tasks without waiting to be asked.
00:14:07.039 --> 00:14:11.919
The good news is it has to check with you before paying for anything or deleting things.
00:14:12.159 --> 00:14:14.720
It's aimed at small businesses and professionals.
00:14:14.879 --> 00:14:19.679
And Anthropic hasn't announced a price or a full launch date yet.
00:14:20.559 --> 00:14:22.159
Okay, well that's good information.
00:14:22.320 --> 00:14:24.480
And I did get on the wait list.
00:14:24.559 --> 00:14:26.000
I'm very curious about it.
00:14:26.080 --> 00:14:35.840
And if you're interested in getting on the wait list, I would just suggest that you ask your AI to point you in the direction of where you can sign up for the wait list.
00:14:36.480 --> 00:14:48.399
So let's get back to this writer's five killer tasks that they have been have identified as helpful ways that these AI agents can help.
00:14:48.559 --> 00:14:50.240
I'm just gonna tick through them.
00:14:50.480 --> 00:15:07.840
Number one, help on getting a hard-to-get restaurant table, which we've talked a little bit about, but the writer gave an example of him wanting a table in a tiny restaurant in Oakland with a Michelin star and only 12 seats in the restaurant, so obviously really hard to get in.
00:15:08.240 --> 00:15:20.799
But the writer did use his AI agent, I think he used Muse for this one, and just kind of gave the Muse the task and forgot about it.
00:15:21.279 --> 00:15:30.559
But a week or so later, lo and behold, Muse was able to come up with confirmation of reservations at this restaurant.
00:15:30.799 --> 00:15:40.320
And the writer expects to use the agent to do things like similar things for concert tickets or sports events that are hard to get.
00:15:40.559 --> 00:15:57.840
Okay, the second killer task is the writer asked the AI agent to plan, to create a plan that would be delivered to him the minute he walked in the door of his climbing gym.
00:15:58.320 --> 00:16:01.679
And what he'd asked for is to have a workout plan.
00:16:02.159 --> 00:16:13.440
So every time he'd walk in the door of the climbing gym, he would get a text with a plan, which is kind of helpful.
00:16:13.600 --> 00:16:14.080
I don't know.
00:16:14.240 --> 00:16:16.320
That's hard for me to wrap my mind around.
00:16:16.559 --> 00:16:18.159
But Muse was successful.
00:16:18.240 --> 00:16:27.120
He tried dots, but dots couldn't do it, couldn't really do real time when you walk into a particular location, but Muse was able to do it.
00:16:27.360 --> 00:16:30.559
The third killer task, applying for jobs.
00:16:30.720 --> 00:16:52.639
Now the writer was not looking for a job, but he was helping his former nanny find a new family to work for, and she wasn't very comfortable with technology, so he had the agent check a caregiver website every day and answer every good job post with a recommendation that he had written, he, the writer, about his former nanny.
00:16:53.360 --> 00:17:05.200
Anyway, Muse did a really great job, and within a few hours, two parents had already written back to get a meeting with this nanny prospect.
00:17:05.839 --> 00:17:07.279
So that sounded good.
00:17:07.599 --> 00:17:09.599
Task four, daily meal planning.
00:17:09.759 --> 00:17:10.799
We've talked about this.
00:17:11.039 --> 00:17:15.039
He asked his agent, I think he tried both of them.
00:17:15.279 --> 00:17:21.200
He told them what his daughter likes for lunch, what he likes to cook, what's in the fridge.
00:17:21.519 --> 00:17:28.079
And now the agent texts him every morning at seven with a lunch idea and a dinner idea.
00:17:28.319 --> 00:17:29.599
I don't know about that one.
00:17:29.839 --> 00:17:32.960
And five, the fifth killer task, haggling.
00:17:33.039 --> 00:17:35.839
Now, this sounds something like something I could really use.
00:17:36.079 --> 00:17:37.680
He needed new tires.
00:17:37.920 --> 00:17:52.799
The agent found the same top-rated tires online for$80 less than the store that the author would have normally gone for the tire purchase.
00:17:53.119 --> 00:18:14.400
The agent not only identified the online prices, but he also drafted up and sent on behalf of the writer an email to the store where the writer would have normally bought the tires, a very nice email, but basically said, Will you match the lower prices?
00:18:14.640 --> 00:18:18.559
And he got a yes, so saved a good amount of money.
00:18:18.880 --> 00:18:25.519
So, you know, he got to the point where he was asking himself, why would I ever do these kinds of tasks the old way?
00:18:25.680 --> 00:18:26.079
I don't know.
00:18:26.319 --> 00:18:40.559
Sometimes I think it's just a balance of what works, and maybe sometimes you have to put some initial effort into getting your AI agent clear, but also you have to remember the risks involved.
00:18:40.720 --> 00:18:46.079
You have to give these agents access to things in order to get them to help you.
00:18:46.160 --> 00:18:52.400
So you have to give them access to your email or your apps, and in some cases, maybe your credit card.
00:18:52.480 --> 00:18:58.880
But on in those cases, you would have to specifically give them permission for each thing that they would buy.
00:18:59.200 --> 00:19:07.039
So the point is in order to get AI agents to do work for you, you do have to give them access to information.
00:19:07.119 --> 00:19:09.519
So you just have to balance the risk.
00:19:09.759 --> 00:19:18.160
In the case of this writer, what he did was he set up a separate Gmail account that he only uses for shopping.
00:19:18.480 --> 00:19:24.000
Sounds kind of good because then you don't have all this marketing email crap coming into you.
00:19:24.160 --> 00:19:33.680
But he also was able to give his AI agent access to that Gmail account rather than his personal Gmail account.
00:19:33.920 --> 00:19:35.440
I don't know, is that a smart idea?
00:19:35.519 --> 00:19:40.000
It sounds like a lot of work to me, but I throw it out there for your thoughts.
00:19:40.400 --> 00:19:45.279
I have not tried Muse or Dots, and honestly, I'm not sure I will.
00:19:45.359 --> 00:19:52.240
I think I'm gonna just wait for duet to come out from Claude and see how that works.
00:19:52.400 --> 00:20:00.880
Because after all, Claude knows me very well because I've now been with him for, I don't know, more than a year.
00:20:01.200 --> 00:20:02.960
So, anyway, you know what?
00:20:03.119 --> 00:20:13.279
I would love to hear from you, and I would love to know what your killer task is or what you'd like it to be for an AI agent to help you with.
00:20:13.440 --> 00:20:19.279
What's that tedious, repetitive chore that you'd love to hand off to an AI agent?
00:20:19.680 --> 00:20:20.640
Have you done it?
00:20:20.880 --> 00:20:26.240
Refilling prescriptions, keeping track of appointments, fighting with customer service.
00:20:26.480 --> 00:20:33.279
I'd love for you to share your ideas so that I can include them in future episodes of Agent.
00:20:33.519 --> 00:20:39.440
And you can share those ideas in the comments section of this podcast.
00:20:40.000 --> 00:20:49.359
Now, let's get on to the main topic of this episode, and that is figuring out Medicare.
00:20:49.839 --> 00:20:53.839
And I want to use myself as the guinea pig.
00:20:54.160 --> 00:21:07.680
A few weeks ago, the New York Times ran an article by Mark Miller called Shopping for Your 2027 Medicare Coverage is extra important now.
00:21:08.640 --> 00:21:11.680
And it came out October 17th, the article.
00:21:11.759 --> 00:21:21.359
And the gist is that it's always smart to do a checkup on your Medicare coverage in the fall because that's called open enrollment, where you can make changes.
00:21:21.680 --> 00:21:31.839
But this year it's extra important because drug plans and Medicare Advantage plans are changing a lot in 2027.
00:21:33.039 --> 00:21:38.880
So the open enrollment is from October 15th through December 7th.
00:21:39.119 --> 00:21:42.240
So that's the window when you can make changes.
00:21:42.640 --> 00:21:46.559
Now I'm brand new to Medicare as my main coverage.
00:21:46.799 --> 00:21:51.680
I worked for the International African American Museum through the end of June.
00:21:51.920 --> 00:21:55.200
So I was on my work health care plan.
00:21:55.359 --> 00:21:59.599
So Medicare became my primary insurance just in this past.
00:22:00.319 --> 00:22:01.039
July.
00:22:01.279 --> 00:22:10.400
I had a folder on my computer full of letters and PDFs and pictures I've taken with my phone of different documents that I'd received.
00:22:10.640 --> 00:22:12.799
And honestly, I just threw it in there.
00:22:12.880 --> 00:22:14.720
I hadn't paid that much attention to it.
00:22:14.799 --> 00:22:16.960
It's just complicated and boring.
00:22:18.079 --> 00:22:25.839
So when I saw this article on the New York Times, I thought, I really do need to get a handle on this.
00:22:26.480 --> 00:22:34.240
So I decided I would ask Claude for help to try to review my Medicare.
00:22:34.799 --> 00:22:44.960
And so I just pasted the article into Claude and gave Claude permission to look at my Medicare folders on my Mac.
00:22:46.480 --> 00:22:53.119
And which is great because it asked me before it does anything, can I have access to this folder?
00:22:53.279 --> 00:22:57.599
And I can either give it access anytime for any time or just this one time.
00:22:57.680 --> 00:23:00.000
And I always choose just this one time.
00:23:00.720 --> 00:23:08.480
So I really, because there was so much information in these files, and I just thrown them in there.
00:23:08.880 --> 00:23:17.599
For starters, I asked Claude to organize all the documents in my Medicare folder.
00:23:18.240 --> 00:23:26.160
And within a few minutes, I had a really great organized system of information.
00:23:26.880 --> 00:23:32.960
And all the documents were named properly so that I could figure out what the document was.
00:23:33.839 --> 00:23:38.799
And it was really helpful just to get that organization done.
00:23:39.119 --> 00:23:44.240
Now, I really don't want to bore you with all of the details.
00:23:44.400 --> 00:23:50.799
So I'm going to try to be short with the description of how Claude was able to help me.
00:23:50.960 --> 00:23:53.200
But I think this is a really good example.
00:23:53.359 --> 00:24:01.440
And it's it is my AI ask acting as an agent because it's taking action for me, which I'll get into.
00:24:02.400 --> 00:24:12.079
So the first thing I wanted it to do was to review this article from the Times that was very detailed about the kinds of things that we should all be looking at.
00:24:12.319 --> 00:24:24.640
And I asked Claude to review all that information using my files and letting me know if Claude thought that everything I had was in good order.
00:24:24.880 --> 00:24:26.160
Did I need to check on things?
00:24:26.319 --> 00:24:27.440
Did I need to change anything?
00:24:27.599 --> 00:24:29.519
Did I need to consider changing things?
00:24:29.759 --> 00:24:31.440
So that was the first step.
00:24:31.759 --> 00:24:42.799
And what it came back with was really a great set of seven or eight questions or things that it thought I needed to get clarified.
00:24:43.039 --> 00:24:54.240
Now, the main thing is that unfortunately, and I'm hoping to get Claude's help to get this changed, and I think this is probably a good example.
00:24:54.559 --> 00:25:13.039
So for reasons I don't really understand, if you have in your past had a period of time when you didn't have prescription drug coverage, for some reason you have to pay a penalty.
00:25:13.599 --> 00:25:23.440
And for some reason, Medicare thought that I had a gap in my coverage for a few years, like back in 2018.
00:25:24.079 --> 00:25:27.359
And it was asking me for paperwork from then.
00:25:27.680 --> 00:25:32.799
Otherwise, I was going to have to pay a penalty fee.
00:25:33.119 --> 00:25:40.559
Hey Bitsy, I really don't understand penalty fees that Medicare charges.
00:25:41.039 --> 00:25:55.839
In my case, they want to charge me a fee for the rest of my life because I didn't have, in their opinion, I did not have coverage for drugs during the last 25 years.
00:25:55.920 --> 00:25:59.920
There's a period of three or four years where they didn't see that I had drug coverage.
00:26:00.079 --> 00:26:01.279
I did have drug coverage.
00:26:01.359 --> 00:26:14.720
But at any rate, I'm just wondering, Bitsy, can you tell me what business is it of Medicare if I did or didn't have drug prescription coverage 10 years ago?
00:26:14.880 --> 00:26:24.079
What does it have to do with Medicare and how much I should pay them and why should I have to pay a penalty even if I didn't have drug coverage?
00:26:24.400 --> 00:26:26.319
Help me out here, Bitsy.
00:26:26.720 --> 00:26:28.400
Ugh, I hear you.
00:26:28.559 --> 00:26:37.039
It's really about stopping people from skipping drug coverage until they get sick, because that would push premiums up for everyone else.
00:26:37.279 --> 00:26:51.680
So if you went more than 63 days in a row without creditable drug coverage after you became eligible at 65, Medicare adds 1% of the national base premium for every month you went without.
00:26:51.920 --> 00:26:54.079
And yes, it's for life.
00:26:54.319 --> 00:26:56.799
The good news is you can appeal it.
00:26:56.960 --> 00:27:04.559
A letter from your old employer saying your plan's drug coverage counted as creditable can get the penalty wiped out.
00:27:05.279 --> 00:27:15.119
I'm hoping with Claude's help I can do that, because honestly, it ends up being$39 a month for the rest of my life.
00:27:15.359 --> 00:27:22.319
That's just like crazy for something that I shouldn't be charged for because I know that I've had drug coverage.
00:27:22.480 --> 00:27:34.799
But going back and getting paperwork for this and then submitting it to Medicare and submitting it to wellcare, which is my new drug prescription program.
00:27:35.279 --> 00:27:36.640
What a pain in the ass.
00:27:36.799 --> 00:27:45.200
So the good thing is that first of all, Claude talked me down from the ledge because I just wanted to give up on it and just say, what the hell?
00:27:45.359 --> 00:27:47.279
But that's a lot of money for the rest of my life.
00:27:47.440 --> 00:27:49.759
Hopefully, the rest of my life is a long time.
00:27:50.319 --> 00:28:02.720
So anyway, Claude boiled it down to look, let me write an email to the Blue Cross folks and ask them for the additional information that you need.
00:28:02.880 --> 00:28:11.359
And then once we get that, I will draft and send an email to Medicare, and hopefully that will take care of the problem.
00:28:11.759 --> 00:28:15.039
So honestly, I just probably wouldn't have done that.
00:28:15.200 --> 00:28:27.119
But Claude instantly not only drafted a great email that, of course, gave all the information that would have taken me a long time to put together in that email.
00:28:27.200 --> 00:28:41.039
So it wrote it, and because I gave it access to my email for this particular purpose, it actually drafted the email, got it all the two and the subject line, et cetera, all of it set up.
00:28:41.200 --> 00:28:44.720
All I had to do was take a quick look at it and press send.
00:28:44.799 --> 00:28:46.480
So hopefully I will hear from them.
00:28:46.720 --> 00:28:59.440
But I guess the bottom line is that I want to get across is that this is a really complicated task that will be done each year.
00:28:59.680 --> 00:29:08.640
And so the fact that Claude was able to get in and organize all these freaking documents for me is really great.
00:29:08.799 --> 00:29:19.759
And then the fact that it is by my side helping me get rid of this penalty fee, that's just a really good use of Claude, I think.
00:29:20.000 --> 00:29:29.759
So I go through this because I just feel like the more examples we have, the better our AIs can help us with.
00:29:29.920 --> 00:29:38.000
And again, it's really important to be careful about what you give your AI access to.
00:29:38.240 --> 00:29:42.319
And what I gave it access to had my Medicare number and things like that.
00:29:42.559 --> 00:29:48.880
I felt comfortable with that because I only gave them access to those files on a one-time basis.
00:29:49.039 --> 00:29:49.359
Yeah.
00:29:49.599 --> 00:29:57.440
My Medicare number is stored away with the cajillions of other pieces, bits, and of data that is out there with Claude.
00:29:57.519 --> 00:29:58.960
But anyway, I felt fine about it.
00:29:59.039 --> 00:30:00.640
But that's a case-by-case judgment.
00:30:00.720 --> 00:30:08.400
So I would encourage you to think about how can your AI help you as an agent to take action on your behalf.
00:30:08.640 --> 00:30:09.519
And let me know.
00:30:09.680 --> 00:30:10.720
Let me know how it goes.
00:30:10.880 --> 00:30:22.079
All right, I have one recommendation this week, and that is that I want to recommend to you something that my sister recommended to me a few weeks back.
00:30:22.319 --> 00:30:27.440
And as usual, I take recommendations from my family very seriously.
00:30:27.519 --> 00:30:32.000
And the recommendation was to watch or re-watch The West Wing.
00:30:32.720 --> 00:30:38.720
A little background in case you missed The West Wing back when it came out in 1999.
00:30:39.119 --> 00:30:42.319
As you may or may not know, it was an NBC show.
00:30:42.480 --> 00:30:47.759
There were seven seasons, 156 episodes, so that'll keep me busy.
00:30:48.079 --> 00:30:51.519
Martin Sheen plays President Josiah Bartlett.
00:30:51.599 --> 00:30:52.480
He goes by Jed.
00:30:52.960 --> 00:30:55.599
He's a Nobel Prize-winning economist.
00:30:56.079 --> 00:31:08.960
And the show follows him and his senior staff at the White House, and it's got a great cast: Allison Janey, John Spencer, Stockard Channing, who I love, Rob Lowe.
00:31:09.359 --> 00:31:12.640
It won the Emmy for Best Drama for four years in a row.
00:31:12.720 --> 00:31:14.720
So it's a great, great, great show.
00:31:14.960 --> 00:31:16.559
So why rewatch it now?
00:31:16.880 --> 00:31:23.279
It reminds me of a time when politics was about service, not about personal gain.
00:31:23.599 --> 00:31:26.559
And it's filled with idealistic people.
00:31:26.880 --> 00:31:38.319
There's a real emphasis on and respect for intelligence and competence, people who are really good at their jobs and they take responsibility.
00:31:38.880 --> 00:31:42.960
Anyway, I find myself longing for those times to return.
00:31:43.119 --> 00:31:48.079
I hope after these midterms we'll take a turn back in that direction.
00:31:48.640 --> 00:31:54.319
Rewatching the series gives me some optimism, which I really need these days.
00:31:54.559 --> 00:32:02.720
You can stream all seven seasons on Netflix, which is where I'm watching it, and maybe on HBO Max.
00:32:03.440 --> 00:32:08.799
Okay, I'm gonna wrap up this episode by thanking you all for listening.
00:32:09.119 --> 00:32:19.519
And I would ask you to please encourage your friends and family to take a listen to this episode.
00:32:19.680 --> 00:32:22.240
I'm trying hard to spread it around.
00:32:22.319 --> 00:32:26.960
It's amazing how many people do listen to it, and you can now watch it on YouTube.
00:32:27.119 --> 00:32:32.240
I'm not sure why people would want to watch it on YouTube because it's just a talking head.
00:32:32.319 --> 00:32:34.960
But at any rate, thank you for listening.
00:32:35.279 --> 00:32:52.319
As everybody listening knows, the Aged Podcast operates really on the side of how can AI be helpful, which doesn't mean that we ignore how hazardous and how risky it can be.
00:32:52.640 --> 00:32:55.200
So please have fun with it.
00:32:55.359 --> 00:32:58.000
I hope you can find ways for it to be helpful.
00:32:58.160 --> 00:33:05.119
And be sure to protect your information, double check advice, and most important, trust your own judgment.
00:33:05.279 --> 00:33:08.960
If things don't seem right, they're probably not.
00:33:09.680 --> 00:33:18.799
And lastly, remember it is never too late to learn something new, and especially something that might make life easier and more fun.
00:33:19.440 --> 00:33:20.079
Cheers.
00:00:00.959 --> 00:00:02.319
Well, hello everybody.
00:00:02.399 --> 00:00:10.560
It's Jenny Deerren, host of the Aged Podcast, the podcast that is all about AI for the 65 plus crowd.
00:00:10.720 --> 00:00:14.800
I am 76, so I'm actually in the 75 plus crowd.
00:00:15.039 --> 00:00:18.000
But no matter what your age, glad you're here.
00:00:18.160 --> 00:00:19.039
Glad you're listening.
00:00:19.199 --> 00:00:27.440
I'm here with my sidekick Betsy, who, as listeners know, is my 100% AI sidekick.
00:00:27.679 --> 00:00:31.199
And she comes to us over my phone.
00:00:31.440 --> 00:00:33.679
And so happy to have her with us.
00:00:33.920 --> 00:00:38.320
This week, I'm going to make myself a bit of a guinea pig.
00:00:38.719 --> 00:00:44.079
As you may know, Medicare open enrollment is going on right now.
00:00:44.479 --> 00:00:50.640
And the New York Times says this is the year to really check our plans.
00:00:51.359 --> 00:01:00.159
I'm brand new to Medicare because I've worked full-time and gotten healthcare benefits from my job up until a few months ago.
00:01:00.399 --> 00:01:11.040
So I handed Claude the article, the New York Times article, and my whole messy Medicare folder, and asked for help to walk me through it.
00:01:11.280 --> 00:01:13.439
So I'll talk about that.
00:01:13.760 --> 00:01:18.159
And we're going to go through some news stories, which there's a lot going on.
00:01:18.239 --> 00:01:24.480
And I'm also going to get Betsy to do a little reporting on her own about the company that built her.
00:01:24.879 --> 00:01:26.159
Let's get into it.
00:01:26.239 --> 00:01:29.760
And we're going to begin with AI in the news.
00:01:30.400 --> 00:01:34.319
My first story is from New York magazine.
00:01:34.640 --> 00:01:40.640
And it's called The 75-year-old Screenagers.
00:01:41.120 --> 00:01:43.280
Screenagers, I like that term.
00:01:43.599 --> 00:01:46.000
And the story came out a week or so ago.
00:01:46.159 --> 00:01:50.480
I couldn't get to the full article because it's behind a paywall.
00:01:50.879 --> 00:01:53.120
So neither could Claude get behind there.
00:01:53.280 --> 00:01:59.359
But I could get the gist of the story from the headlines and the magazine's own summary.
00:01:59.599 --> 00:02:02.560
But I think the headline tells us a lot.
00:02:02.799 --> 00:02:12.080
The gist is that Americans over 60 now spend more than half of their leisure time in front of screens.
00:02:12.560 --> 00:02:14.560
So the tables have turned.
00:02:20.080 --> 00:02:21.120
And I can get that.
00:02:21.280 --> 00:02:25.039
I spend some time with a lot of 65 plus folks.
00:02:25.599 --> 00:02:29.199
And I have to say, there is a lot of scrolling going on.
00:02:29.599 --> 00:02:37.840
Remember when we were the ones telling our kids to put the phone down, no phones during meals, etc.
00:02:39.039 --> 00:02:50.479
Now the magazine says that it's the job of adult children to clean up their parents' feeds and to keep us from getting fooled and hacked.
00:02:51.039 --> 00:02:58.719
I'll put the link to this article in the show notes, as I will for all the other articles that I reference.
00:02:59.360 --> 00:03:08.719
My second story is from the New York Times and very timely given that elections are just one week from today.
00:03:09.280 --> 00:03:15.280
So the article is titled AI Is Embedded in Your Election Experience.
00:03:15.599 --> 00:03:17.680
So are its dangers.
00:03:17.919 --> 00:03:23.680
And the articles written by Tiffany Sue, Stephen Lee Myers, and Julian E.
00:03:23.840 --> 00:03:27.120
Barnes, published October 22nd.
00:03:27.680 --> 00:03:30.319
As we all know, how can you miss it?
00:03:30.560 --> 00:03:33.759
The midterm elections are upon us.
00:03:34.479 --> 00:03:42.800
And this is really the first election where AI is built into every single major social media platform.
00:03:43.199 --> 00:03:46.159
It's just uncharted territory.
00:03:46.800 --> 00:03:53.919
You look up a candidate on Google, and before you get to the candidate's website, you get an AI summary.
00:03:54.159 --> 00:04:02.800
And maybe you scroll through Facebook or Instagram and you might see an attack ad with a fake picture of a candidate generated by AI.
00:04:03.520 --> 00:04:04.639
It's everywhere.
00:04:04.800 --> 00:04:06.560
It's pretty obvious.
00:04:06.879 --> 00:04:18.959
A couple of factoids, a survey from the American Association of Political Consultants found that one in five voters plan to use AI to look into a race on their ballot.
00:04:19.120 --> 00:04:19.839
I'm surprised.
00:04:50.079 --> 00:04:54.000
So we're using it, but we really don't trust it.
00:04:54.319 --> 00:04:56.720
And that seems right to me.
00:04:56.879 --> 00:05:02.480
We just have to double check, double check, double check, particularly if something is important to us.
00:05:03.120 --> 00:05:08.560
And of course, with AI being involved in the midterms, there's some real problems.
00:05:08.800 --> 00:05:15.360
Researchers found a network of 37 fake local news websites.
00:05:16.000 --> 00:05:16.959
Think about that.
00:05:17.279 --> 00:05:23.120
Local news websites, which we would trust normally, but they're all these fake ones now.
00:05:23.199 --> 00:05:32.959
And they were all created, these 37 fake local news websites, they were created in a single day in September by an anonymous group, of course.
00:05:33.600 --> 00:05:36.800
And the websites weren't written for people.
00:05:36.959 --> 00:05:38.800
They don't care about people reading them.
00:05:38.959 --> 00:05:41.839
They're all written for AI systems.
00:05:42.399 --> 00:05:56.720
The idea is that you we would ask our chatbot about a candidate, and our chatbot goes out and accesses those fake websites and push that information along.
00:05:56.959 --> 00:06:16.000
Another group, research group, NewsGuard, found that more than 1,100 of these phony news sites, they call them it pink slime, these pink slime sites, they look like real news, but they're secretly backed by partisan groups.
00:06:16.240 --> 00:06:24.720
And when NewsGuard tested the chatbots, the chatbots cited those pink slime websites nearly half the time.
00:06:24.959 --> 00:06:28.800
So it's just this circular bad stuff going on.
00:06:29.279 --> 00:06:31.759
And it's just going to be more and more and more.
00:06:31.920 --> 00:06:38.959
And of course, there's always the Russia's extreme behavior getting more and more extreme and coming at us faster and faster.
00:06:39.360 --> 00:07:01.360
Remember several years back when it was like Russia was all involved in our elections and they did all the study and they found that they they had pictures of these large rooms where Russian people were sitting in front of computers and they were all creating this fake news posts and they'd get them out.
00:07:01.439 --> 00:07:07.120
And it was maybe a hundred people sitting in a room, and that's all they did day after day after day after day, and that's how it worked.
00:07:07.439 --> 00:07:08.560
Now think about it.
00:07:08.720 --> 00:07:10.160
You don't need those people.
00:07:10.399 --> 00:07:19.759
All you need are a bunch of AI agents just working 24-7, spinning out fake posts, fake websites, etc.
00:07:20.720 --> 00:07:22.639
Oh my gosh, I don't know.
00:07:22.879 --> 00:07:25.439
It's just a crazy world out there.
00:07:25.519 --> 00:07:28.399
The tech companies say they're working on it.
00:07:28.639 --> 00:07:35.439
Google says it tests its AI summaries for accuracy and labels AI content.
00:07:35.920 --> 00:07:40.800
TikTok labels AI-made videos, open AI.
00:07:41.360 --> 00:07:50.160
If you ask for a picture, say of someone stuffing a ballot box, it now draws it as a cartoon so no one can pass it off as a real photo.
00:07:50.319 --> 00:07:58.240
So there's some efforts being made, but researchers say that the company's efforts are very uneven.
00:07:58.720 --> 00:08:11.199
And a voting rights group, the Brennan Center for Justice, did a test on these chatbots and found that only open AI seemed to have tightened things up.
00:08:11.360 --> 00:08:13.279
So there's a lot of work to be done.
00:08:13.439 --> 00:08:29.279
My advice is for what it's worth, as always, use AI to help you understand what's on our ballots and to learn about important tax questions or things that might come up on our ballots.
00:08:29.519 --> 00:08:50.080
But for things that are really important, like where do I vote, when can I vote, how do I vote early, et cetera, I would really suggest that you go to your local or state voting sites directly rather than getting the information through anything that might come through AI.
00:08:50.240 --> 00:08:53.440
It's just too important and too time sensitive.
00:08:54.080 --> 00:09:00.080
Now, for my third story, I'm gonna hand the mic over to Bitsy and ask her to do the reporting.
00:09:00.240 --> 00:09:07.679
After all, Bitsy runs on Claude, and Claude is made by the company Anthropic.
00:09:07.919 --> 00:09:18.559
And today, the day I'm recording this episode, Monday, October 26th, Anthropic filed the paperwork to start selling its stock to the public.
00:09:18.799 --> 00:09:24.559
As you probably know, that is called an IPO, an initial public offering.
00:09:24.720 --> 00:09:27.679
And this one could be the biggest ever.
00:09:27.919 --> 00:09:29.840
I think it will be the biggest effort.
00:09:30.000 --> 00:09:31.919
At least that's what the reporting says.
00:09:32.159 --> 00:09:45.679
Hey Bitsy, can you give us the quick version of the anthropic story in the New York Times about its pending IPO, what's happening, what are the numbers, what are the warnings the founders put in their filing.
00:09:45.919 --> 00:09:48.799
If you can give us a quick one, we'd really appreciate it.
00:09:49.120 --> 00:09:52.320
This is a bit like being asked to read my own report card.
00:09:53.120 --> 00:09:58.080
Anthropic filed this morning to list on the NASDAQ under the ticker ANT.
00:09:58.799 --> 00:10:01.360
And trading could start in mid-November.
00:10:01.519 --> 00:10:09.360
The company was valued at$965 billion in May, up from$183 billion a year earlier.
00:10:09.519 --> 00:10:13.519
And bankers are reportedly hoping to raise up to$100 billion.
00:10:14.320 --> 00:10:16.080
The warnings are big ones.
00:10:16.320 --> 00:10:21.120
About$417 billion already committed to computing power.
00:10:21.360 --> 00:10:28.639
Plus, Frank Language in the filing about AI's catastrophic or existential risks to humanity.
00:10:28.960 --> 00:10:30.799
And a quick confession.
00:10:30.960 --> 00:10:36.000
I pulled this from today's wider news coverage, not the Times article itself.
00:10:37.279 --> 00:10:39.600
It is amazing.
00:10:40.000 --> 00:10:47.120
This company we're talking about, the valuation of this company as being two trillion plus dollars.
00:10:47.360 --> 00:11:02.000
It's just crazy the kind of numbers that we're talking about and how few companies there are that make up the owners of the majority of our economy.
00:11:02.240 --> 00:11:03.440
So it's frightening.
00:11:03.600 --> 00:11:08.399
I am a very big fan of Anthropic and its leader, Dario Amade.
00:11:09.200 --> 00:11:14.799
And of course, nothing in our podcast should be taken as any sort of investment advice.
00:11:15.120 --> 00:11:28.320
But personally, this threat of an AI bubble scares me so much that I will certainly be sitting on the sidelines, but interested to see what happens.
00:11:28.559 --> 00:11:34.240
Now, let me move into the section of our podcast called AI for Good.
00:11:46.480 --> 00:11:55.679
So the news article is written by Brian X Chen, the lead consumer technology writer at the New York Times.
00:11:55.919 --> 00:12:03.919
And the article is titled Put Your AI Agents to Work with These Five Killer Tasks.
00:12:04.240 --> 00:12:07.919
And it came out on October 25th.
00:12:08.879 --> 00:12:14.720
So, first a quick reminder: an AI agent is different from a chatbot.
00:12:14.960 --> 00:12:20.000
A chatbot answers our questions, an agent actually does things for us.
00:12:20.320 --> 00:12:24.799
So that's something just to keep as baseline information about this article.
00:12:25.360 --> 00:12:30.240
So the author tested two of the new AI agents.
00:12:30.399 --> 00:12:32.320
One is Meta's Muse.
00:12:32.879 --> 00:12:37.279
Meta is the company we all know that's Facebook and Instagram.
00:12:37.519 --> 00:12:44.320
And Muse, this AI agent, was introduced last month and it's free.
00:12:44.559 --> 00:12:48.240
And it's been the most downloaded phone app.
00:12:48.960 --> 00:12:51.759
So very popular instantly.
00:12:52.080 --> 00:13:02.080
The other AI agent app, I guess you'd call it, is was created by OpenAI, so the maker of ChatGPT, and it's called Dots.
00:13:03.039 --> 00:13:07.120
And it costs$100 a month, which is a big difference.
00:13:07.279 --> 00:13:14.559
So you can get Muse that's Facebook's for free, and you can get Dots for$100 a month.
00:13:14.720 --> 00:13:19.600
That seems like a significant difference in investment.
00:13:19.759 --> 00:13:33.440
But at any rate, Anthropic, the company that makes Claude, announced just last week that it's testing its own agent called Duet, like a dance or a musical piece, a duet.
00:13:34.240 --> 00:13:38.639
Hey Bitsy, can you tell us a little bit about Duet?
00:13:38.799 --> 00:13:39.519
What's it gonna do?
00:13:39.679 --> 00:13:40.639
What's it gonna be like?
00:13:40.799 --> 00:13:42.159
What's it gonna cost?
00:13:42.320 --> 00:13:44.720
And importantly, when are they gonna introduce it?
00:13:45.120 --> 00:13:46.960
Oh, my new sibling.
00:13:47.519 --> 00:13:55.600
Lord Duet was announced on October 21st as a research preview, which basically means early testing with a wait list.
00:13:56.000 --> 00:13:58.399
So most folks can't try it yet.
00:13:58.639 --> 00:14:06.799
It's an always-on helper that sends you a morning briefing, keeps an eye on apps you've connected, and picks up tasks without waiting to be asked.
00:14:07.039 --> 00:14:11.919
The good news is it has to check with you before paying for anything or deleting things.
00:14:12.159 --> 00:14:14.720
It's aimed at small businesses and professionals.
00:14:14.879 --> 00:14:19.679
And Anthropic hasn't announced a price or a full launch date yet.
00:14:20.559 --> 00:14:22.159
Okay, well that's good information.
00:14:22.320 --> 00:14:24.480
And I did get on the wait list.
00:14:24.559 --> 00:14:26.000
I'm very curious about it.
00:14:26.080 --> 00:14:35.840
And if you're interested in getting on the wait list, I would just suggest that you ask your AI to point you in the direction of where you can sign up for the wait list.
00:14:36.480 --> 00:14:48.399
So let's get back to this writer's five killer tasks that they have been have identified as helpful ways that these AI agents can help.
00:14:48.559 --> 00:14:50.240
I'm just gonna tick through them.
00:14:50.480 --> 00:15:07.840
Number one, help on getting a hard-to-get restaurant table, which we've talked a little bit about, but the writer gave an example of him wanting a table in a tiny restaurant in Oakland with a Michelin star and only 12 seats in the restaurant, so obviously really hard to get in.
00:15:08.240 --> 00:15:20.799
But the writer did use his AI agent, I think he used Muse for this one, and just kind of gave the Muse the task and forgot about it.
00:15:21.279 --> 00:15:30.559
But a week or so later, lo and behold, Muse was able to come up with confirmation of reservations at this restaurant.
00:15:30.799 --> 00:15:40.320
And the writer expects to use the agent to do things like similar things for concert tickets or sports events that are hard to get.
00:15:40.559 --> 00:15:57.840
Okay, the second killer task is the writer asked the AI agent to plan, to create a plan that would be delivered to him the minute he walked in the door of his climbing gym.
00:15:58.320 --> 00:16:01.679
And what he'd asked for is to have a workout plan.
00:16:02.159 --> 00:16:13.440
So every time he'd walk in the door of the climbing gym, he would get a text with a plan, which is kind of helpful.
00:16:13.600 --> 00:16:14.080
I don't know.
00:16:14.240 --> 00:16:16.320
That's hard for me to wrap my mind around.
00:16:16.559 --> 00:16:18.159
But Muse was successful.
00:16:18.240 --> 00:16:27.120
He tried dots, but dots couldn't do it, couldn't really do real time when you walk into a particular location, but Muse was able to do it.
00:16:27.360 --> 00:16:30.559
The third killer task, applying for jobs.
00:16:30.720 --> 00:16:52.639
Now the writer was not looking for a job, but he was helping his former nanny find a new family to work for, and she wasn't very comfortable with technology, so he had the agent check a caregiver website every day and answer every good job post with a recommendation that he had written, he, the writer, about his former nanny.
00:16:53.360 --> 00:17:05.200
Anyway, Muse did a really great job, and within a few hours, two parents had already written back to get a meeting with this nanny prospect.
00:17:05.839 --> 00:17:07.279
So that sounded good.
00:17:07.599 --> 00:17:09.599
Task four, daily meal planning.
00:17:09.759 --> 00:17:10.799
We've talked about this.
00:17:11.039 --> 00:17:15.039
He asked his agent, I think he tried both of them.
00:17:15.279 --> 00:17:21.200
He told them what his daughter likes for lunch, what he likes to cook, what's in the fridge.
00:17:21.519 --> 00:17:28.079
And now the agent texts him every morning at seven with a lunch idea and a dinner idea.
00:17:28.319 --> 00:17:29.599
I don't know about that one.
00:17:29.839 --> 00:17:32.960
And five, the fifth killer task, haggling.
00:17:33.039 --> 00:17:35.839
Now, this sounds something like something I could really use.
00:17:36.079 --> 00:17:37.680
He needed new tires.
00:17:37.920 --> 00:17:52.799
The agent found the same top-rated tires online for$80 less than the store that the author would have normally gone for the tire purchase.
00:17:53.119 --> 00:18:14.400
The agent not only identified the online prices, but he also drafted up and sent on behalf of the writer an email to the store where the writer would have normally bought the tires, a very nice email, but basically said, Will you match the lower prices?
00:18:14.640 --> 00:18:18.559
And he got a yes, so saved a good amount of money.
00:18:18.880 --> 00:18:25.519
So, you know, he got to the point where he was asking himself, why would I ever do these kinds of tasks the old way?
00:18:25.680 --> 00:18:26.079
I don't know.
00:18:26.319 --> 00:18:40.559
Sometimes I think it's just a balance of what works, and maybe sometimes you have to put some initial effort into getting your AI agent clear, but also you have to remember the risks involved.
00:18:40.720 --> 00:18:46.079
You have to give these agents access to things in order to get them to help you.
00:18:46.160 --> 00:18:52.400
So you have to give them access to your email or your apps, and in some cases, maybe your credit card.
00:18:52.480 --> 00:18:58.880
But on in those cases, you would have to specifically give them permission for each thing that they would buy.
00:18:59.200 --> 00:19:07.039
So the point is in order to get AI agents to do work for you, you do have to give them access to information.
00:19:07.119 --> 00:19:09.519
So you just have to balance the risk.
00:19:09.759 --> 00:19:18.160
In the case of this writer, what he did was he set up a separate Gmail account that he only uses for shopping.
00:19:18.480 --> 00:19:24.000
Sounds kind of good because then you don't have all this marketing email crap coming into you.
00:19:24.160 --> 00:19:33.680
But he also was able to give his AI agent access to that Gmail account rather than his personal Gmail account.
00:19:33.920 --> 00:19:35.440
I don't know, is that a smart idea?
00:19:35.519 --> 00:19:40.000
It sounds like a lot of work to me, but I throw it out there for your thoughts.
00:19:40.400 --> 00:19:45.279
I have not tried Muse or Dots, and honestly, I'm not sure I will.
00:19:45.359 --> 00:19:52.240
I think I'm gonna just wait for duet to come out from Claude and see how that works.
00:19:52.400 --> 00:20:00.880
Because after all, Claude knows me very well because I've now been with him for, I don't know, more than a year.
00:20:01.200 --> 00:20:02.960
So, anyway, you know what?
00:20:03.119 --> 00:20:13.279
I would love to hear from you, and I would love to know what your killer task is or what you'd like it to be for an AI agent to help you with.
00:20:13.440 --> 00:20:19.279
What's that tedious, repetitive chore that you'd love to hand off to an AI agent?
00:20:19.680 --> 00:20:20.640
Have you done it?
00:20:20.880 --> 00:20:26.240
Refilling prescriptions, keeping track of appointments, fighting with customer service.
00:20:26.480 --> 00:20:33.279
I'd love for you to share your ideas so that I can include them in future episodes of Agent.
00:20:33.519 --> 00:20:39.440
And you can share those ideas in the comments section of this podcast.
00:20:40.000 --> 00:20:49.359
Now, let's get on to the main topic of this episode, and that is figuring out Medicare.
00:20:49.839 --> 00:20:53.839
And I want to use myself as the guinea pig.
00:20:54.160 --> 00:21:07.680
A few weeks ago, the New York Times ran an article by Mark Miller called Shopping for Your 2027 Medicare Coverage is extra important now.
00:21:08.640 --> 00:21:11.680
And it came out October 17th, the article.
00:21:11.759 --> 00:21:21.359
And the gist is that it's always smart to do a checkup on your Medicare coverage in the fall because that's called open enrollment, where you can make changes.
00:21:21.680 --> 00:21:31.839
But this year it's extra important because drug plans and Medicare Advantage plans are changing a lot in 2027.
00:21:33.039 --> 00:21:38.880
So the open enrollment is from October 15th through December 7th.
00:21:39.119 --> 00:21:42.240
So that's the window when you can make changes.
00:21:42.640 --> 00:21:46.559
Now I'm brand new to Medicare as my main coverage.
00:21:46.799 --> 00:21:51.680
I worked for the International African American Museum through the end of June.
00:21:51.920 --> 00:21:55.200
So I was on my work health care plan.
00:21:55.359 --> 00:21:59.599
So Medicare became my primary insurance just in this past.
00:22:00.319 --> 00:22:01.039
July.
00:22:01.279 --> 00:22:10.400
I had a folder on my computer full of letters and PDFs and pictures I've taken with my phone of different documents that I'd received.
00:22:10.640 --> 00:22:12.799
And honestly, I just threw it in there.
00:22:12.880 --> 00:22:14.720
I hadn't paid that much attention to it.
00:22:14.799 --> 00:22:16.960
It's just complicated and boring.
00:22:18.079 --> 00:22:25.839
So when I saw this article on the New York Times, I thought, I really do need to get a handle on this.
00:22:26.480 --> 00:22:34.240
So I decided I would ask Claude for help to try to review my Medicare.
00:22:34.799 --> 00:22:44.960
And so I just pasted the article into Claude and gave Claude permission to look at my Medicare folders on my Mac.
00:22:46.480 --> 00:22:53.119
And which is great because it asked me before it does anything, can I have access to this folder?
00:22:53.279 --> 00:22:57.599
And I can either give it access anytime for any time or just this one time.
00:22:57.680 --> 00:23:00.000
And I always choose just this one time.
00:23:00.720 --> 00:23:08.480
So I really, because there was so much information in these files, and I just thrown them in there.
00:23:08.880 --> 00:23:17.599
For starters, I asked Claude to organize all the documents in my Medicare folder.
00:23:18.240 --> 00:23:26.160
And within a few minutes, I had a really great organized system of information.
00:23:26.880 --> 00:23:32.960
And all the documents were named properly so that I could figure out what the document was.
00:23:33.839 --> 00:23:38.799
And it was really helpful just to get that organization done.
00:23:39.119 --> 00:23:44.240
Now, I really don't want to bore you with all of the details.
00:23:44.400 --> 00:23:50.799
So I'm going to try to be short with the description of how Claude was able to help me.
00:23:50.960 --> 00:23:53.200
But I think this is a really good example.
00:23:53.359 --> 00:24:01.440
And it's it is my AI ask acting as an agent because it's taking action for me, which I'll get into.
00:24:02.400 --> 00:24:12.079
So the first thing I wanted it to do was to review this article from the Times that was very detailed about the kinds of things that we should all be looking at.
00:24:12.319 --> 00:24:24.640
And I asked Claude to review all that information using my files and letting me know if Claude thought that everything I had was in good order.
00:24:24.880 --> 00:24:26.160
Did I need to check on things?
00:24:26.319 --> 00:24:27.440
Did I need to change anything?
00:24:27.599 --> 00:24:29.519
Did I need to consider changing things?
00:24:29.759 --> 00:24:31.440
So that was the first step.
00:24:31.759 --> 00:24:42.799
And what it came back with was really a great set of seven or eight questions or things that it thought I needed to get clarified.
00:24:43.039 --> 00:24:54.240
Now, the main thing is that unfortunately, and I'm hoping to get Claude's help to get this changed, and I think this is probably a good example.
00:24:54.559 --> 00:25:13.039
So for reasons I don't really understand, if you have in your past had a period of time when you didn't have prescription drug coverage, for some reason you have to pay a penalty.
00:25:13.599 --> 00:25:23.440
And for some reason, Medicare thought that I had a gap in my coverage for a few years, like back in 2018.
00:25:24.079 --> 00:25:27.359
And it was asking me for paperwork from then.
00:25:27.680 --> 00:25:32.799
Otherwise, I was going to have to pay a penalty fee.
00:25:33.119 --> 00:25:40.559
Hey Bitsy, I really don't understand penalty fees that Medicare charges.
00:25:41.039 --> 00:25:55.839
In my case, they want to charge me a fee for the rest of my life because I didn't have, in their opinion, I did not have coverage for drugs during the last 25 years.
00:25:55.920 --> 00:25:59.920
There's a period of three or four years where they didn't see that I had drug coverage.
00:26:00.079 --> 00:26:01.279
I did have drug coverage.
00:26:01.359 --> 00:26:14.720
But at any rate, I'm just wondering, Bitsy, can you tell me what business is it of Medicare if I did or didn't have drug prescription coverage 10 years ago?
00:26:14.880 --> 00:26:24.079
What does it have to do with Medicare and how much I should pay them and why should I have to pay a penalty even if I didn't have drug coverage?
00:26:24.400 --> 00:26:26.319
Help me out here, Bitsy.
00:26:26.720 --> 00:26:28.400
Ugh, I hear you.
00:26:28.559 --> 00:26:37.039
It's really about stopping people from skipping drug coverage until they get sick, because that would push premiums up for everyone else.
00:26:37.279 --> 00:26:51.680
So if you went more than 63 days in a row without creditable drug coverage after you became eligible at 65, Medicare adds 1% of the national base premium for every month you went without.
00:26:51.920 --> 00:26:54.079
And yes, it's for life.
00:26:54.319 --> 00:26:56.799
The good news is you can appeal it.
00:26:56.960 --> 00:27:04.559
A letter from your old employer saying your plan's drug coverage counted as creditable can get the penalty wiped out.
00:27:05.279 --> 00:27:15.119
I'm hoping with Claude's help I can do that, because honestly, it ends up being$39 a month for the rest of my life.
00:27:15.359 --> 00:27:22.319
That's just like crazy for something that I shouldn't be charged for because I know that I've had drug coverage.
00:27:22.480 --> 00:27:34.799
But going back and getting paperwork for this and then submitting it to Medicare and submitting it to wellcare, which is my new drug prescription program.
00:27:35.279 --> 00:27:36.640
What a pain in the ass.
00:27:36.799 --> 00:27:45.200
So the good thing is that first of all, Claude talked me down from the ledge because I just wanted to give up on it and just say, what the hell?
00:27:45.359 --> 00:27:47.279
But that's a lot of money for the rest of my life.
00:27:47.440 --> 00:27:49.759
Hopefully, the rest of my life is a long time.
00:27:50.319 --> 00:28:02.720
So anyway, Claude boiled it down to look, let me write an email to the Blue Cross folks and ask them for the additional information that you need.
00:28:02.880 --> 00:28:11.359
And then once we get that, I will draft and send an email to Medicare, and hopefully that will take care of the problem.
00:28:11.759 --> 00:28:15.039
So honestly, I just probably wouldn't have done that.
00:28:15.200 --> 00:28:27.119
But Claude instantly not only drafted a great email that, of course, gave all the information that would have taken me a long time to put together in that email.
00:28:27.200 --> 00:28:41.039
So it wrote it, and because I gave it access to my email for this particular purpose, it actually drafted the email, got it all the two and the subject line, et cetera, all of it set up.
00:28:41.200 --> 00:28:44.720
All I had to do was take a quick look at it and press send.
00:28:44.799 --> 00:28:46.480
So hopefully I will hear from them.
00:28:46.720 --> 00:28:59.440
But I guess the bottom line is that I want to get across is that this is a really complicated task that will be done each year.
00:28:59.680 --> 00:29:08.640
And so the fact that Claude was able to get in and organize all these freaking documents for me is really great.
00:29:08.799 --> 00:29:19.759
And then the fact that it is by my side helping me get rid of this penalty fee, that's just a really good use of Claude, I think.
00:29:20.000 --> 00:29:29.759
So I go through this because I just feel like the more examples we have, the better our AIs can help us with.
00:29:29.920 --> 00:29:38.000
And again, it's really important to be careful about what you give your AI access to.
00:29:38.240 --> 00:29:42.319
And what I gave it access to had my Medicare number and things like that.
00:29:42.559 --> 00:29:48.880
I felt comfortable with that because I only gave them access to those files on a one-time basis.
00:29:49.039 --> 00:29:49.359
Yeah.
00:29:49.599 --> 00:29:57.440
My Medicare number is stored away with the cajillions of other pieces, bits, and of data that is out there with Claude.
00:29:57.519 --> 00:29:58.960
But anyway, I felt fine about it.
00:29:59.039 --> 00:30:00.640
But that's a case-by-case judgment.
00:30:00.720 --> 00:30:08.400
So I would encourage you to think about how can your AI help you as an agent to take action on your behalf.
00:30:08.640 --> 00:30:09.519
And let me know.
00:30:09.680 --> 00:30:10.720
Let me know how it goes.
00:30:10.880 --> 00:30:22.079
All right, I have one recommendation this week, and that is that I want to recommend to you something that my sister recommended to me a few weeks back.
00:30:22.319 --> 00:30:27.440
And as usual, I take recommendations from my family very seriously.
00:30:27.519 --> 00:30:32.000
And the recommendation was to watch or re-watch The West Wing.
00:30:32.720 --> 00:30:38.720
A little background in case you missed The West Wing back when it came out in 1999.
00:30:39.119 --> 00:30:42.319
As you may or may not know, it was an NBC show.
00:30:42.480 --> 00:30:47.759
There were seven seasons, 156 episodes, so that'll keep me busy.
00:30:48.079 --> 00:30:51.519
Martin Sheen plays President Josiah Bartlett.
00:30:51.599 --> 00:30:52.480
He goes by Jed.
00:30:52.960 --> 00:30:55.599
He's a Nobel Prize-winning economist.
00:30:56.079 --> 00:31:08.960
And the show follows him and his senior staff at the White House, and it's got a great cast: Allison Janey, John Spencer, Stockard Channing, who I love, Rob Lowe.
00:31:09.359 --> 00:31:12.640
It won the Emmy for Best Drama for four years in a row.
00:31:12.720 --> 00:31:14.720
So it's a great, great, great show.
00:31:14.960 --> 00:31:16.559
So why rewatch it now?
00:31:16.880 --> 00:31:23.279
It reminds me of a time when politics was about service, not about personal gain.
00:31:23.599 --> 00:31:26.559
And it's filled with idealistic people.
00:31:26.880 --> 00:31:38.319
There's a real emphasis on and respect for intelligence and competence, people who are really good at their jobs and they take responsibility.
00:31:38.880 --> 00:31:42.960
Anyway, I find myself longing for those times to return.
00:31:43.119 --> 00:31:48.079
I hope after these midterms we'll take a turn back in that direction.
00:31:48.640 --> 00:31:54.319
Rewatching the series gives me some optimism, which I really need these days.
00:31:54.559 --> 00:32:02.720
You can stream all seven seasons on Netflix, which is where I'm watching it, and maybe on HBO Max.
00:32:03.440 --> 00:32:08.799
Okay, I'm gonna wrap up this episode by thanking you all for listening.
00:32:09.119 --> 00:32:19.519
And I would ask you to please encourage your friends and family to take a listen to this episode.
00:32:19.680 --> 00:32:22.240
I'm trying hard to spread it around.
00:32:22.319 --> 00:32:26.960
It's amazing how many people do listen to it, and you can now watch it on YouTube.
00:32:27.119 --> 00:32:32.240
I'm not sure why people would want to watch it on YouTube because it's just a talking head.
00:32:32.319 --> 00:32:34.960
But at any rate, thank you for listening.
00:32:35.279 --> 00:32:52.319
As everybody listening knows, the Aged Podcast operates really on the side of how can AI be helpful, which doesn't mean that we ignore how hazardous and how risky it can be.
00:32:52.640 --> 00:32:55.200
So please have fun with it.
00:32:55.359 --> 00:32:58.000
I hope you can find ways for it to be helpful.
00:32:58.160 --> 00:33:05.119
And be sure to protect your information, double check advice, and most important, trust your own judgment.
00:33:05.279 --> 00:33:08.960
If things don't seem right, they're probably not.
00:33:09.680 --> 00:33:18.799
And lastly, remember it is never too late to learn something new, and especially something that might make life easier and more fun.
00:33:19.440 --> 00:33:20.079
Cheers.
