E8: GPT4 - AI Unleashed w/ ChinaTalk Podcast

Potential of GPT4, US v. China 'racing dynamics' and Runway Gen-2

Hey Cogs,

Today we’ll discuss:

• GPT-4

• US v. China 'racing dynamics'

• Runway Gen-2

---

GPT4

Thankfully, our friend

Ben Tossell

of

Ben's Bites

curated the best examples of GPT-4. It’s a great and inspiring thread.

Ben Tossell @bentossell

All the best examples of GPT-4, from OpenAI:

US v. China 'racing dynamics'

The race between the US and China in AI has been a topic of discussion for quite some time. With billions of dollars at stake and the geopolitical dynamics involved, the competition has intensified. China has made it a strategic goal to create superior generalized models in AI, and the US sees this as a critical technology that must be advanced. The competitive spirit between the two nations has created a racing dynamic that some feel is problematic, while others view it as a positive force for progress. In recent months, however, there has been a modestly positive update in the race dynamics. The cost of inference has dropped significantly, making universal basic intelligence more affordable and accessible. While OpenAI dominates the market and has integrated its product everywhere, there are still enough big companies like Google, Amazon, and Apple that see the strategic importance of AI and will continue to compete in the field. As a result, it's expected that we will see a narrow field of contenders in the West, but it remains to be seen how the competition will play out in China.

---

Thank you Omneky for sponsoring The Cognitive Revolution. Omneky is an omnichannel creative generation platform that lets you launch hundreds of thousands of ad iterations that actually work, customized across all platforms, with a click of a button. Omneky combines generative AI and real-time advertising data, to generate personalized experiences at scale.

---

Runway Gen-2

Runway, a next-generation video editing solution powered by AI and ML, released Gen-2, a multi-modal diffusion model capable of processing text, images, and video to generate high-quality videos.

Runway @runwayml

Generate videos with nothing but words. If you can say it, now you can see it. Introducing, Text to Video. With Gen-2. Learn more at research.runwayml.com/gen2

While alternatives on HuggingFace may not match the scale and quality of Gen-2, the model's capabilities have already surpassed expectations. This is a significant achievement since video processing has always been more complicated than image or text processing. With the success of Gen-2, the idea of a singularity event in 2030 seems increasingly plausible.

In the near future, it may be possible for users to craft a feature-length film script tailored to their preferences using GPT and then generate an entire movie through Runway without even reading it. The age of personal content creation is quickly approaching, and with models like Gen-2, the possibilities are endless. These advancements are exciting for AI engineers, designers, and researchers who continue to push the boundaries of what is possible with AI and multi-modal models.

To dive deeper into multi-modals, you can listen to our interview with BLIP/BLIP2 founder, Junnan Li and Dongxu Li and our interview with Playground AI’ Suhail Doshi is a great primer on the advancements in text-to-image.

---

On the next episode of The Cognitive Revolution, we have Flo Crivello, founder of Teamflow. The episode drops today.

Until next time.