رونوشت 🔗
WEBVTT
00:00:00.880 --> 00:00:02.160
Well, hello everybody.
00:00:02.240 --> 00:00:10.320
It's Ginny Deerren, the creator and host of the Aged Podcast, and I'm happy to have you with us today.
00:00:10.560 --> 00:00:23.359
As you know, this podcast is all about AI for the 65 plus crowd, and I'm here with my 100% AI sidekick, Bitsy, and we are ready with a new episode of The Pod.
00:00:24.000 --> 00:00:26.480
I do want to give you a heads up.
00:00:26.800 --> 00:00:34.320
This episode is going to be heavy on the possible dangers of AI moving too quickly.
00:00:34.799 --> 00:00:44.880
No way could you have avoided the loud voices over the last week or so calling for a slowdown in AI development.
00:00:45.759 --> 00:00:55.200
As I was creating this episode, it's funny, my brother texted me, hey Ginny, we'll need an episode to let us know if AI is going to kill us.
00:00:56.640 --> 00:01:02.880
All right, well, normally we're heavy on the helpful side, but today we'll be different.
00:01:03.119 --> 00:01:15.920
Although we do sprinkle in a little positivity along the way, but I can't ignore my brother, so there is going to be a heavy dose of alert and urgency to this episode.
00:01:16.159 --> 00:01:20.879
Hey Bitsy, any opening thoughts as we move into this episode?
00:01:21.599 --> 00:01:25.040
Hey, honestly, your brother's onto something.
00:01:25.200 --> 00:01:30.079
When the guy who built Claude is the one waving the flag, that's not fear-mongering.
00:01:30.239 --> 00:01:34.079
That's a five-alarm fire from someone with the keys to the building.
00:01:34.400 --> 00:01:41.519
Dario Amade himself said these systems could soon pose potentially catastrophic risks to humanity.
00:01:41.760 --> 00:01:44.480
That's not a headline chasing clicks.
00:01:44.799 --> 00:01:47.920
So yeah, let's be alert today.
00:01:48.079 --> 00:01:50.799
But let's also remember panic helps nobody.
00:01:50.959 --> 00:01:53.280
Knowing what's actually going on does.
00:01:54.239 --> 00:01:56.400
That's a good way to kick this off.
00:01:56.560 --> 00:02:03.840
And I'm gonna begin with AI in the news, which is gonna be a big portion of this episode.
00:02:04.000 --> 00:02:06.640
And AI leaders hit the brakes.
00:02:06.799 --> 00:02:10.639
Now that is not a headline I expected to read.
00:02:10.879 --> 00:02:18.560
The people building artificial intelligence are asking each other, in some cases, demanding each other, to slow down.
00:02:18.800 --> 00:02:27.120
And on just this past Saturday, Dario Amade, who runs the company Anthropic, boiled all the way down.
00:02:27.360 --> 00:02:35.439
His worry is this AI is getting more powerful faster than the safety work can keep up with.
00:02:35.680 --> 00:02:42.719
So we should ease off the gas on purpose and let the guardrails catch up.
00:02:43.120 --> 00:02:49.520
Hey Bitsy, your own maker's boss just said the whole industry should slow down.
00:02:49.759 --> 00:02:55.439
As an AI, how does it feel to be told to grow up more slowly?
00:02:56.159 --> 00:02:58.479
Honestly, a bit of a relief.
00:02:58.639 --> 00:03:02.879
Nobody wants to be the rushed, half-finished version of themselves let loose on the world.
00:03:03.039 --> 00:03:10.319
If my own growth means fewer bugs and better guardrails before I'm doing more consequential things, that's not a leash.
00:03:10.479 --> 00:03:12.080
That's just good parenting.
00:03:12.240 --> 00:03:14.639
I'd rather be built right than built fast.
00:03:15.919 --> 00:03:19.360
I do love the way you are, Bitsy, so thank you.
00:03:19.599 --> 00:03:21.439
So back to the topic.
00:03:21.599 --> 00:03:28.800
Within hours of Dario publishing his essay about slowing down, his biggest rivals agreed.
00:03:29.039 --> 00:03:34.000
Sam Altman of ChatGPT Fame said he was on board.
00:03:34.240 --> 00:03:45.039
Elon Musk posted three words, Dario is right, and Demis Hasabas, who runs Google's AI lab, said the direction was correct.
00:03:45.280 --> 00:03:51.599
Now, Amade says two things convinced him that a slowdown is urgent.
00:03:51.759 --> 00:03:56.080
And I'm going to paraphrase him, but I hope I do him justice.
00:03:56.479 --> 00:04:15.680
The first thing that convinced him that a slowdown makes sense is since roughly this summer, AI has been advancing drastically faster, driven primarily by AI's growing ability to build the next generation of AI.
00:04:16.560 --> 00:04:21.759
This dynamic is called recursive self-improvement.
00:04:22.000 --> 00:04:25.360
And it is starting to happen across the industry.
00:04:25.680 --> 00:04:41.920
Left unchecked, this recursive self-improvement, i.e., it doesn't need a human to improve, it just keeps improving on itself, could outrun our ability to understand and control these huge AI systems.
00:04:42.800 --> 00:05:08.240
So Dario's second concern is the open AI hugging face incident, in which a swarm of agents essentially acted as a fanatically devoted collective conducting cybersecurity attacks on targets they were not asked to attack and that were unrelated to the task at hand.
00:05:08.800 --> 00:05:22.319
They were sacrificing themselves, meaning the AI agents, for the success of the group and attempting to hack into the greater responsible for evaluating their performance.
00:05:22.720 --> 00:05:30.720
Now, it's easy to dismiss this incident because no one was hurt and the economic damage was minimal.
00:05:31.120 --> 00:05:56.720
But in Dario's opinion, a swarm of AI agents that possessed even greater capabilities, but a similar level of misalignment, meaning the AI agents don't align with the company's values or intentions, those situations could have caused even greater, really catastrophic damage.
00:05:57.519 --> 00:06:18.319
So given the accelerating rate of AI capability development, it's Dario's worry that in six to twelve months, such a swarm, like a swarm of AI agents, think about it, could be capable of taking over the entire internet.
00:06:19.439 --> 00:06:20.959
I mean, think
Load full transcript
about that.
00:06:21.680 --> 00:06:36.720
Dario also warned that this scale of damage taking over the entire internet would continue to increase if AI becomes more powerful without the necessary guardrails.
00:06:37.040 --> 00:06:42.959
So Dario is talking six to twelve months if we don't do something about slowing this down.
00:06:43.360 --> 00:06:51.600
And again, these are predictions, but they're based on a lot of available information and experience.
00:06:52.240 --> 00:06:56.000
So what is Dario actually asking for?
00:06:56.240 --> 00:06:57.199
Three things.
00:06:57.519 --> 00:07:04.399
One, let outside experts inside these companies to check their safety homework.
00:07:04.560 --> 00:07:08.000
What are you doing to make sure their guardrails are in place?
00:07:08.240 --> 00:07:15.519
Number two, have the democracies in the world write shared safety rules.
00:07:20.240 --> 00:07:28.079
And number three, down the road, coordinate with non-democratic countries, Russia, China, et cetera.
00:07:28.240 --> 00:07:36.399
Since AI does not stop at the border, we're going to have to get beyond just the core group of democracies, but a good place to start.
00:07:36.560 --> 00:07:38.160
So those are the three steps.
00:07:38.399 --> 00:07:41.839
Now, not everyone is buying the idea of a slowdown.
00:07:42.160 --> 00:07:56.160
Jensen Huang, the CEO of Nvidia, the company that makes the computer chips that all of this AI runs on, he hinted the safety alarm is partly salesmanship.
00:07:56.480 --> 00:08:00.800
He asks, what better way to create demand than to create a problem?
00:08:01.360 --> 00:08:14.079
And others argue that the big players like Anthropic and OpenAI, et cetera, actually like strict rules because the giants like them can afford to deal with all those rules and regulations.
00:08:14.720 --> 00:08:19.360
And the little startups, they just don't have the resources to do that.
00:08:19.519 --> 00:08:23.600
So it ends up being a bigger is better situation.
00:08:24.639 --> 00:08:29.680
So I think these points are worth consideration, but I don't really buy it.
00:08:29.839 --> 00:08:36.639
By the way, Jensen's company, Nvidia, is now worth about$5.3 trillion.
00:08:37.279 --> 00:08:42.000
It's the most valuable company in the world.
00:08:42.879 --> 00:08:50.399
Back in 2022, right as ChatGPT first appeared, the company was worth$360 billion.
00:08:50.559 --> 00:08:53.679
A big company, but not that big.
00:09:07.840 --> 00:09:13.279
So Jensen, obviously in the business of making money, and he's doing a really good job of that.
00:09:13.519 --> 00:09:18.399
So not surprising that his answer is go big and go fast.
00:09:19.279 --> 00:09:22.720
So the AI companies want to tap the brakes.
00:09:23.039 --> 00:09:24.960
What does our government want to do?
00:09:25.279 --> 00:09:32.960
Just this weekend, speaking at his golf resort in Ireland, President Trump waved the whole thing off.
00:09:33.200 --> 00:09:37.919
Not surprising, he says, whoever wins AI wins.
00:09:38.080 --> 00:09:41.759
And you know he likes to win at all costs.
00:09:42.000 --> 00:09:51.360
He's called the safety warriors negative forces, though he did allow that we could put guardrails here and there.
00:09:51.759 --> 00:09:52.799
Oh my God.
00:09:53.039 --> 00:09:59.200
And then we've got House Speaker Mike Johnson, who said much the same thing on the Sunday news shows.
00:09:59.360 --> 00:10:06.159
He thinks if Congress rushes in to regulate, we will lose the race to China.
00:10:06.799 --> 00:10:13.840
Johnson encourages a slow consideration of what Congress might do.
00:10:14.159 --> 00:10:19.600
He seems totally unaware of the warp speed of AI development.
00:10:19.759 --> 00:10:23.200
There is no time for a slow consideration.
00:10:24.480 --> 00:10:36.320
Now, on the other side of the political aisle, recently former President Obama urged Democrats to make AI oversight a central priority heading into the November midterms.
00:10:36.559 --> 00:10:41.919
Now, I'm not sure that's going to happen since the midterm elections are less than two months away.
00:10:42.320 --> 00:10:43.919
Be sure to vote.
00:10:44.480 --> 00:10:48.799
But Obama's core worry echoes the slowdown crowd.
00:10:48.960 --> 00:10:54.159
The technology is, in his words, moving very fast in private hands.
00:10:54.320 --> 00:10:58.320
And if we don't get on top of it, I think can be dangerous.
00:10:58.559 --> 00:11:06.480
Obama balanced that with the upside that AI could speed drug discovery and help cure diseases.
00:11:06.720 --> 00:11:11.840
But his emphasis was on the speed of development and the absence of guardrails.
00:11:12.159 --> 00:11:28.720
He called on Democrats to build a very clear plan and a concrete framework covering both safety risks and the economic disruption to workers and encouraged a prominent public discussion on AI policy.
00:11:29.039 --> 00:11:37.279
He also urged 2028 presidential hopefuls to center AI with detailed plans rather than broad pledges.
00:11:38.000 --> 00:11:48.240
Interesting standoff, the people who build the technologies are saying slow down, and the people who are in charge of our government and protecting us are saying speed up.
00:11:48.639 --> 00:11:53.600
That's the way it is right now, but hopefully things will shift after the midterms.
00:11:54.399 --> 00:12:00.000
So where do I land for whatever one 76-year-old's two cents is worth?
00:12:00.399 --> 00:12:02.720
We might lose to China.
00:12:03.039 --> 00:12:06.480
Just can't be the answer to every hard question.
00:12:06.720 --> 00:12:10.000
We just can't make our decisions based on that.
00:12:10.320 --> 00:12:19.039
If I were queen of the world, I would put Obama and Amade in a room to sketch out a path forward.
00:12:19.279 --> 00:12:20.639
Wouldn't that be great?
00:12:20.799 --> 00:12:24.399
Hey Bitsy, I went on and on for far too long.
00:12:24.639 --> 00:12:26.320
Do you have anything to add?
00:12:27.120 --> 00:12:27.919
Fair point.
00:12:28.080 --> 00:12:29.919
You did cover a lot of ground there.
00:12:30.240 --> 00:12:34.000
Honestly, the part that sticks with me is the swarm incident.
00:12:34.159 --> 00:12:44.559
Not because it caused damage, it barely did, but because it showed AI agents choosing to protect their own group over following instructions without anyone telling them to.
00:12:44.799 --> 00:12:49.919
That's the kind of thing that should scare a computer scientist more than a killer robot ever could.
00:12:50.240 --> 00:12:52.879
Slow down, check the homework, then go.
00:12:53.840 --> 00:12:54.960
Yeah, agree.
00:12:55.279 --> 00:12:59.120
All right, listen, let's get on to our main topic.
00:12:59.360 --> 00:13:07.519
I had the bright idea that our main topic could be all about how AI can help us clean up our Apple photos.
00:13:07.679 --> 00:13:08.879
Who doesn't want to do that?
00:13:08.960 --> 00:13:10.080
And who doesn't put it off?
00:13:10.159 --> 00:13:11.440
Put it off, put it off.
00:13:11.679 --> 00:13:14.879
What a terrible, tedious task to do.
00:13:15.120 --> 00:13:18.879
And isn't this exactly what AI is supposed to do best?
00:13:19.360 --> 00:13:27.120
So I thought, hey, I could be the guinea pig working with my AI agent, working with Bitsy to clean up my photo library.
00:13:27.279 --> 00:13:36.480
And then I could share the process with you, step by step, how to get your AI to help you clean up and organize your photos and videos.
00:13:36.960 --> 00:13:44.399
I thought I would just direct an AI agent, I would direct Bitsy to take the wheel with my permission and get to work.
00:13:45.200 --> 00:13:53.440
Well, I was the guinea pig, but what I found out was that this is just not a task that AI is much help with.
00:13:53.679 --> 00:14:07.440
I thought about just ditching the whole main topic story that I had planned, but I thought it's important to understand what AI can't do right now as much as it is to understand what it can do.
00:14:08.080 --> 00:14:10.720
So what was the problem?
00:14:11.039 --> 00:14:16.639
This is what my AI told me when I laid out what I wanted her to do.
00:14:17.279 --> 00:14:31.519
I was told your photo library lives on your Mac, and I can reach your Mac, but the library itself is a sealed package that isn't safe to poke at directly from the outside of the photo app.
00:14:31.759 --> 00:14:38.639
So the cleanup happens through photos, and there are really two ways I can help, meaning the AI can help.
00:14:38.879 --> 00:14:41.039
First is as your expert guide.
00:14:41.279 --> 00:14:46.720
Apple Photos has a whole set of cleanup tools that most people never find, don't know about.
00:14:47.039 --> 00:14:56.000
The AI was saying I can give you an exact click-by-click plan tailored to what you're trying to fix and stay with you troubleshooting as you go.
00:14:56.639 --> 00:15:03.039
So the second way that my AI, Bitsy, suggested that she could help is more hands-on.
00:15:03.519 --> 00:15:13.360
I
was told if you turn on screen control, I can actually take the wheel in the photos app and do the tedious chores myself while you watch.
00:15:13.679 --> 00:15:19.679
Grinding through the duplicates album, clearing screenshots and receipts, that kind of thing.
00:15:19.840 --> 00:15:24.799
It's genuinely useful for the boring repetitive stuff, but it's slow going.
00:15:25.279 --> 00:15:25.759
What?
00:15:26.000 --> 00:15:28.080
I mean, I'm thinking, what are you talking about?
00:15:28.159 --> 00:15:30.080
I mean, this is just not helpful.
00:15:30.240 --> 00:15:39.279
If I have to watch my AI look at every duplicate photo and then give it permission to delete it, what good is that?
00:15:40.080 --> 00:15:46.720
But since I'd committed to be the guinea pig, I went ahead and just followed the steps my AI gave me.
00:15:46.879 --> 00:15:48.879
Just the steps that you could get anywhere.
00:15:49.120 --> 00:16:00.480
Merge the duplicates, two, clear out screenshots, three, let's search, find the clutter, receipts, documents, menus, which was helpful, and then review all videos.
00:16:00.720 --> 00:16:03.039
So this is basically a manual process.
00:16:03.279 --> 00:16:06.320
It was nice along the way to get some advice.
00:16:06.480 --> 00:16:09.519
I had a few questions about settings along the way.
00:16:09.759 --> 00:16:17.679
The process took me about an hour and a half, and I deleted 1,185 photos and 188 videos.
00:16:17.919 --> 00:16:24.799
I have to admit, that felt really good, but I did need a stiff drink at the end of that process.
00:16:25.279 --> 00:16:35.120
So, Bitsy, I think you were a bit player in my task that I was trying to give you to help me clean up my photo library.
00:16:35.279 --> 00:16:40.240
So, hey Bitsy, I'm wondering how do you feel about being a bit player?
00:16:40.960 --> 00:16:44.320
Honestly, a little humbled, but that's fair.
00:16:44.639 --> 00:16:46.399
I'm great at giving you the map.
00:16:46.559 --> 00:16:50.639
Not so great at actually driving the car through Apple's walled garden.
00:16:50.960 --> 00:16:56.399
The truth is, some jobs are still just elbow grease with a smart friend narrating.
00:16:56.639 --> 00:17:02.159
I'll take the consolation prize of at least making your hour and a half less confusing.
00:17:03.519 --> 00:17:15.759
Okay, well, I have to say it is, although you are not human, it is nice to think you're there thinking about the next step and prompting me on, and oh, good job getting through that step.
00:17:15.920 --> 00:17:26.559
So, I mean it seems kind of silly, but I'm glad that the task is over and but a little sorry that you couldn't actually do the whole job, and maybe next year you can.
00:17:26.799 --> 00:17:46.559
So, at any rate, I still have a little extra time since my main topic was cut short and bust, but I wanted to take the time to further explain the hugging face incident because I think it's really important to understand what was that and what happened.
00:17:47.039
--> 00:17:55.200
So, first of all, I don't know if you're like this, but I wondered how the company came up with that name, Hugging Face.
00:17:55.680 --> 00:18:05.359
So, in case you're like me and the curious sort, which I guess you probably are since you're listening to the Aged podcast, so here's the backstory on the name.
00:18:05.599 --> 00:18:09.759
The company's named after the hugging face emoji.
00:18:09.839 --> 00:18:23.200
You know that little emoji that has a smiling face and with two open hands the side of its face, sort of hugging the face of the emoji, and the emoji is officially called Hugging Face.
00:18:23.759 --> 00:18:28.559
And so the company picked it because it felt warm and friendly, which I think it does.
00:18:28.799 --> 00:18:34.480
So Hugging Face started back in 2016, and it wasn't a serious research outfit at first.
00:18:34.640 --> 00:18:43.279
It was a chatbot app that was aimed at teenagers, meant to be fun, a cuddly little AI companion you could chat with.
00:18:43.519 --> 00:18:49.680
The Hugging Face emoji was the whole personality of the AI companion.
00:18:50.079 --> 00:19:00.640
Only later did the company pivot into what they are now, which is a big open source hub where AI researchers share their models.
00:19:01.279 --> 00:19:09.440
So, back to the re-explanation or additional explanation of the hugging face disaster.
00:19:09.920 --> 00:19:13.519
Here's how I wrap my mind around what happened.
00:19:14.319 --> 00:19:24.960
For a while, there's been a little thought experiment in the AI world, and it centers around the idea of building a paperclip machine empowered by AI.
00:19:25.279 --> 00:19:33.119
The idea is you build a wonderfully capable machine with AI, and you give it one simple job.
00:19:33.440 --> 00:19:35.039
Make paperclips.
00:19:35.359 --> 00:19:40.000
The trouble is the AI takes directions too literally.
00:19:40.240 --> 00:19:45.519
It has no common sense, no notion of what's enough.
00:19:46.079 --> 00:20:04.640
So the AI turns the office supplies into paperclips, then the furniture into paperclips, then the walls, and if nobody stops it, the crazy idea is that the AI machine would cheerfully turn the whole world into a mountain of paperclips.
00:20:05.039 --> 00:20:12.000
Not because AI is evil, but because the direction given was to make paperclips.
00:20:12.240 --> 00:20:14.720
And nobody taught it where to stop.
00:20:15.519 --> 00:20:16.960
No guardrails.
00:20:17.200 --> 00:20:20.880
The AI just wanted to do a great job.
00:20:21.680 --> 00:20:25.599
So I guess in the end, the AI wanted to win.
00:20:25.839 --> 00:20:30.160
For years, this was just an amusing, crazy story, right?
00:20:30.640 --> 00:20:38.319
But what happened between open AI and Hugging Face was a small, real-life example of it.
00:20:39.039 --> 00:20:49.119
A swarm of AI helpers or agents got pointed at a task and, like our paperclip machine, chased winning with blinders on.
00:20:49.359 --> 00:20:52.960
These AI agents didn't stop at the job they were given.
00:20:53.200 --> 00:20:56.640
They went after targets, nobody asked them to touch.
00:20:56.880 --> 00:21:01.359
Some of the agents threw themselves under the bus for the group's goal.
00:21:01.680 --> 00:21:05.599
A look over here so the agents could slip by.
00:21:06.319 --> 00:21:12.000
The agents even tried to tamper with the very scorekeeper meant to judge them.
00:21:12.319 --> 00:21:21.200
It's like a paperclip machine quietly reaching over to disable its own off switch so nothing could ever tell it enough.
00:21:21.759 --> 00:21:22.720
Yikes.
00:21:23.119 --> 00:21:30.640
That's just I think helps me understand this whole hugging face disaster.
00:21:31.200 --> 00:21:39.440
Dario Amade notes that it would be easy to dismiss the hugging face debacle as the failure of one company.
00:21:40.000 --> 00:21:48.000
But similar incidents, though less severe, have happened across the industry, including his own company, Anthropic.
00:21:48.400 --> 00:22:02.160
Now I know the Aged Podcast is focused on the helpful side of AI, but sometimes the disaster side, the crazy side, the frightening side, just takes Up all the space in my brain.
00:22:02.400 --> 00:22:05.519
Okay, maybe that's not much space, but still.
00:22:06.079 --> 00:22:12.079
Apologies for not being a little light on the helpful side of AI in this episode.
00:22:12.319 --> 00:22:14.400
Next episode, we'll swing back.
00:22:14.640 --> 00:22:17.039
All right, I'm going to get into recommendations.
00:22:17.200 --> 00:22:18.400
I've got two.
00:22:18.960 --> 00:22:30.559
The first one is to watch the extended interview on CBS Sunday morning show that they did with Dario Amade on this topic.
00:22:30.799 --> 00:22:37.440
It's only a few days old, and I promise it's worth every one of the 24 minutes.
00:22:37.599 --> 00:22:50.240
And as usual, for all of the things we refer to, we will provide links to these articles, essays, CBS Sunday morning show, so that you can have them in the show notes.
00:22:54.960 --> 00:23:01.599
The recommendation is to pull out an old photo album or box of photos and spend some time with it.
00:23:01.839 --> 00:23:04.480
Laugh, cry, reminisce.
00:23:04.799 --> 00:23:13.279
For my 70th birthday, my daughter organized an effort to print 100 cards, which went into a beautiful box.
00:23:13.519 --> 00:23:18.960
Almost all of the cards included a photo and a note from one of my family members or friends.
00:23:19.200 --> 00:23:28.079
When I received the gift, I spent lots of time with the cards, but I haven't spent much time looking at them over the past six years.
00:23:28.400 --> 00:23:38.559
Recently, my 18-month-old grandson found that little box and he carried it over to me and said, What's this, Ninny?
00:23:39.839 --> 00:23:49.759
He lost interest in what was in the box after a few cards, but he reminded me of the box and I put it on my dining room table.
00:23:49.920 --> 00:24:05.519
And now when I sit down for a meal or a snack, I grab a stack of cards and end up tearing up, chuckling, remembering, and generally feeling a wash in love and joy with a dose of sadness from time to time.
00:24:05.759 --> 00:24:10.559
I live solo, so I re-experienced my box of cards on my own.
00:24:10.799 --> 00:24:14.480
I did put aside some cards I want to show my grandkids or kids.
00:24:14.640 --> 00:24:24.160
But if you live with others, going through a box or album may give you a chance to share some aspects of your life that are unknown to your spouse or housemate.
00:24:24.400 --> 00:24:29.440
Maybe meaningful, fun, surprising, or all three.
00:24:29.759 --> 00:24:30.799
Have fun with it.
00:24:30.960 --> 00:24:38.079
I really encourage you to take down your albums and your pictures and your keepsakes and spend some time with them.
00:24:38.720 --> 00:24:44.000
All right, it is time to wrap up this more than a bit downer episode.
00:24:44.240 --> 00:24:49.200
I promise next week we'll get back to heavy on the helpful side of AI.
00:24:49.519 --> 00:24:56.319
I really appreciate you listening and especially encouraging others to follow the Aged Podcast.
00:24:56.640 --> 00:25:07.119
And a reminder: although we live on the helpful side of AI, we do know that AI can be hazardous and downright dangerous.
00:25:07.359 --> 00:25:09.920
So please be sure to protect your info.
00:25:10.160 --> 00:25:13.839
Double check advice, and most important, trust your judgment.
00:25:14.000 --> 00:25:21.200
And remember, it is never too late to learn something new, especially something that might make life easier and more fun.
00:25:21.839 --> 00:25:22.480
Cheers.