EP44: The Finale: Google Gemini, SimTheory, Is Ilya OK? Predictions for 2024
Dec 8, 2023
auto_awesome
In the final episode of the year, the podcast covers Google's Gemini AI models, their first impressions, and the likely impact on the market. They introduce SimTheory, where listeners can use AI agents mentioned on the show. The podcast also addresses the drama at OpenAI and makes predictions for AI in 2024. They discuss Meta's latest announcements, including the AI Alliance for AI Openness. The speakers express gratitude for their audience and look forward to the next year.
Google's announcement of their new AI technology, Gemini, lacked focus and failed to resonate with the AI community and developers
Google's marketing approach with the Gemini announcement was criticized for its lack of real-world use cases and excessive emphasis on safety measures
OpenAI is facing internal challenges following a recent upheaval, resulting in slower models and increased timeouts
Deep dives
Google's Gemini Announcement and Disappointment
Google recently made a big announcement with their new AI technology, Gemini, which generated a lot of excitement initially. However, upon closer inspection, it was revealed that the demonstrations shown in the video were overly contrived and didn't accurately represent the actual capabilities of the technology. The announcement lacked focus and failed to resonate with the AI community and developers. Google would have been better off highlighting their training methodology and the future capabilities of Gemini rather than presenting misleading demos.
Gemini Models and Limited Access
Google introduced three models with the Gemini announcement: nano, pro, and ultra. Nano is designed for on-device tasks, such as suggested replies, while pro is compared to GPT-3.5 in terms of capabilities. The ultra model is advertised as being better than GPT-4 on various benchmarks, though limited access means it's difficult to verify these claims. The announcement left many users feeling disappointed as access to the models was restricted, region-locked, and lacking API availability.
Google's Bold Claims and Safety Measures
Google's marketing approach with the Gemini announcement failed to impress the AI community, mainly due to the lack of real-world use cases and cherry-picked examples in their videos. The company's emphasis on safety measures and disclaimers for supposedly unsaved prompts came across as excessive. Google could benefit from being more transparent and avoiding overly contrived demos, while focusing on its long-term commitment to AI development.
Google's Future in AI and the Competitive Landscape
Despite Google's claim that Gemini represents the future of AI, it remains to be seen whether they can deliver on their promises. Challenges in scaling the models to meet user expectations and the need for practical applications could hinder Gemini's success. OpenAI, with its better distribution and marketing strategies, continues to maintain a leading position in the AI field. Google's success in the AI space will depend on their ability to gain developer trust, demonstrate long-term commitment, and provide genuine value to users.
OpenAI facing internal issues
OpenAI is facing internal challenges following the recent upheaval, with the issues that caused the blowup not fully resolved. This has led to a lack of focus on technology and scaling, resulting in slower models and increased timeouts. The organization is also grappling with internal politics, which is negatively impacting its progress.
Open source models gaining traction
Open source models are gaining momentum and providing viable alternatives to proprietary models. While no single open source model dominates the market, the advent of smaller, fine-tuned models is making specialized tasks more efficient and accurate. With hardware becoming cheaper, more companies are exploring the use of larger memory size models locally. The rise of open source models is leading to a more diverse AI landscape and may drive the establishment of a model store with specialized models for specific tasks.
In our final episode for the year, we cover the surprise announcement of Google's Gemini AI models and give our first impressions. We road test Gemini Pro on Bard and discuss the likely impact of Gemini on the market and developer ecosystems. Then it's time for our holiday gift: SimTheory. Now you can use AI agents we mention on the show including our virtual girlfriends, Sports Betting with AI and many more! You can even create your own agents to try different models using the same tools we use to prepare for the show. We then discuss if Ilya is OK and the drama at OpenAI. And finally, we make predictions for 2024 and cover some of Meta's latest announcements.
Thanks for watching, listening and all your support through 2023. We really appreciate it and will see you early next year!
CHAPTERS: ===== 00:00 - Google Gemini is Here? Kinda 38:48 - Our Holiday Gift: SimTheory: Virtual Girlfriend, Sports Betting with AI Agents 51:15 - Is Ilya OK? Is GPT-4 Slowness About Cost Reductions? 56:26 - NexusRaven-V2-13B for function calling: is this the future of specialized fine tune models? 1:00:14 - Our Predictions for AI in 2024 1:12:54 - Meta announces AI Alliance for AI Openness + Updates to Meta AI Characters and SeamlessExpressive 1:15:43 - Final thoughts and thank you
