ABOUT THIS EPISODE
In this episode of Human-First AI Marketing, Mike Montague talks with Sherif Higazy, Founder of Megaton AI, a benchmarking and advisory firm for generative media, about one of the hardest questions in AI right now: how do you evaluate which models are actually useful when everything is changing so quickly? Sherif shares what Megaton is seeing across AI video model leaderboards, why Chinese video models are currently pushing the creative frontier, and how marketers can think more clearly about model quality, cost, privacy, and real-world use cases.
The conversation explores the future of AI-generated video, music, animation, visual effects, real-time generation, world models, and immersive experiences that may soon feel closer to a Holodeck than a traditional editing suite. Mike and Sherif also unpack the “attention tax” of AI slop, the role of human authorship, and why the best creative results still come from people who know what they want to say before they ask AI to help make it.
Key Takeaways:
- Keeping up with AI model changes has become nearly impossible without active testing. Sherif explains that constant marketing hype makes it hard to know which models are actually good unless someone is regularly evaluating them.
- Creative AI models are harder to evaluate than text-based LLMs. Unlike text models that can be tested with clear tasks, video and creative models require more subjective judgment around quality, style, and artistic usefulness.
- Chinese AI video models are currently leading the pack. Sherif points to models from Chinese labs, especially ByteDance’s SeeDance 2.5, as some of the strongest options for cinematic AI video generation.
- Open-source and local models matter for companies with sensitive IP. For businesses that cannot upload confidential material to cloud-based tools, locally run models create a safer path to using AI video.
- AI video works best when users build workflows instead of expecting instant magic. Sherif emphasizes that strong results come from developing an eye for quality, learning the tools, and treating AI as part of a creative production process.
- AI is expanding what can be made, especially for teams with smaller budgets. He shares that projects once blocked by the high cost of visual effects, animation, or production may now become financially possible.
- Human authorship still determines whether AI-generated content feels worth watching. Sherif warns that audiences can sense when content lacks care, intention, or a clear creative reason to exist.
- AI slop creates an attention tax. He describes low-effort AI content as a burden on people’s time and attention, especially when it takes longer to consume than it took to make.
- Some of the strongest AI use cases are animation, visual effects, training videos, marketing videos, and music. Sherif sees these areas as practical entry points where AI can help people create things that were previously expensive or out of reach.
- World models may be the next major leap in creative AI. Sherif predicts that tools capable of turning images into explorable 3D spaces could open the door to Holodeck-like experiences, immersive history, gaming, and new storytelling formats.
At Avenue9, we capture your expertise and turn it into content, conversations, and client testimonials, building trust with humans and AI at the same time.
Get your free Human-AI Trust Score report.
Want to be a guest? Send Mike Montague a message on PodMatch here.
