• The Strange Review
  • Posts
  • Did Q*, a new breakthrough in AGI, spark the OpenAI drama last week?

Did Q*, a new breakthrough in AGI, spark the OpenAI drama last week?

Google delays release of Gemini AI, Deepmind and Youtube release gen-AI music model Lyria, Stability announces Stable Video

Author

Tara Tan
November 23, 2023

Well. I don’t think anyone had the OpenAI fall-out on their 2023 bingo card but here we are.

Last Friday afternoon (Nov 17), some unexpected news dropped: OpenAI’s board unceremoniously ousted CEO Sam Altman because they felt like he "was not consistently candid in his communications with the board”. About what? No one quite knows for sure, exactly.

Just a few hours ago, a Reuters article speculates that it might have to do with Q*, a breakthrough in the startup’s search for AGI (artificial general intelligence), autonomous AI with human-level cognition and reasoning abilities. Check out the Reddit Thread on What is Q*?.

Over the weekend, the AI industry experienced whiplash as the board, investors, and CEOs tussled over the future of what is arguably, the most influential company in AI today.

A quick recap:

Nov 17 — Altman is publicly ousted as CEO. Everyone is in shock. Their investors, including Microsoft who invested $13B in OpenAI, were blindsided.

Nov 18 — OpenAI investors push hard for Altman’s return. Altman publicly plots new venture.

Nov 19 — Board negotiations hit a snag. Emmett Shear, the co-founder of video streaming site Twitch, is appointed as interim CEO. (Hint: a few days later, his Twitter bio reads “interim ex-CEO of OpenAI”)

Nov 20 — Altman and Greg Brockman (former President of Open AI) announces that they will join Microsoft to lead a new AI research team. 650 out of 770 Open AI employees publish a letter saying that they might quit en masse unless the startup’s board resigns and reappoints the ousted Altman.

Nov 21 — In a surprise turn of events, Open AI reveals that Altman will return as company’s CEO, and a new board will be appointed, which will include former Salesforce chief executive Bret Taylor, former US Secretary of the Treasury Larry Summers and Quora founder Adam D’Angelo.

All’s well ends well? Only time will tell. Whatever happens next, we will definitely be in for a ride.

My personal takeaways from this series of events:

  • Founders, choose your board (and cofounders) wisely.

  • MSFT CEO Satya Nadella’s moves this weekend was a masterclass in leadership. With a deft hand, he turned a messy situation into a win for MSFT in just 48 hours. Bravo.

  • It seems likely that we will see more breakthroughs in AGI within 18 months. There may be attempts to increase control through regulation.

  • We’ll likely see increasing polarization within the research community (and perhaps, general population) around AGI, open source vs closed source, and the extent which technology and invention should be regulated.

For those who celebrate, happy Thanksgiving. 🦃 

The Long Read

Microsoft CEO Nadella says OpenAI governance needs to change no matter where Altman ends up. He told CNBC’s Jon Fortt  that the governance structure of OpenAI needs to change, three days after the sudden firing of CEO Sam Altman. In his first press interview since Altman’s ouster, Nadella dismissed concerns of long-term damage at OpenAI and said that the critical artificial intelligence research continues as does the partnership with Microsoft. (link)

🔥 Latest news

  • Google delays release of Gemini AI that aims to compete with OpenAI. Google’s company-defining effort to catch up to ChatGPT creator OpenAI is turning out to be harder than expected. Google representatives earlier this year told some cloud customers and business partners they would get access to the company’s new conversational AI, a large language model known as Gemini, by November. But the company recently told them not to expect it until the first quarter of next year, according to two people with direct knowledge. The delay comes at a bad time for Google, whose cloud sales growth has slowed while that of its bigger rival, Microsoft, has accelerated. https://bitly.ws/32UtB 

  • Anthropic seizes the moment, unveiling Claude 2.1 as OpenAI implodes. As OpenAI faces an uncertain future following the surprise ouster of its CEO Sam Altman, competitor Anthropic is seizing the moment to release its own updated LLM, Claude 2.1. The most significant advancement is a 200,000 token context window, allowing Claude to process documents up to 150,000 words or 500 pages long. This enables analysis of entire codebases, lengthy financial reports, research papers and other complex documents. Anthropic has differentiated itself with a rigorous focus on AI safety. https://tinyurl.com/3wbur2s6 

  • Meta brings us a step closer to AI-generated movies. Meta debuted Emu Video, an evolution of its image generation tool, Emu. Given a caption, image or a photo paired with a description, Emu Video can generate a four-second animated clip. Emu Video’s clips can be edited with a complementary AI model called Emu Edit, which was also announced. Users can describe the modifications they want to make to Emu Edit in natural language — e.g. “the same clip, but in slow motion” — and see the changes reflected in a newly generated video. https://tcrn.ch/46cuzaQ 

  • DeepMind and YouTube release Lyria, a gen-AI model for music and Dream Track to build AI tunes. Google DeepMind has announced a new music generation model called Lyria that will work in conjunction with YouTube, and two new toolsets it’s describing as “experiments” built on Lyria: Dream Track lets users create music for YouTube Shorts, and Music AI is a set of tools that it says are aimed at helping with the creative process (for example, building a tune out of a snipped that a creator might hum). Alongside these, DeepMind said it’s adapting SynthID — used to mark AI images — to watermark AI music, too. https://tcrn.ch/47zQwBF 

  • Microsoft releases Orca 2, a pair of small language models that outperform larger counterparts. The research arm of MSFT dropped Orca 2, a pair of small language models that either match or outperform five to ten times larger language models, including Meta’s Llama-2 Chat-70B, when tested on complex reasoning tasks in zero-shot settings. The models come in two sizes, 7 billion and 13 billion parameters, and build on the work done on the original 13B Orca model that demonstrated strong reasoning abilities by imitating step-by-step reasoning traces of bigger, more capable models a few months ago. https://tinyurl.com/2v93b287 

  • Meta dissolves responsible AI division amid restructuring. Meta has reportedly disbanded its division responsible for regulating its AI ventures as they are developed and deployed. According to a report, many team members of Meta’s responsible AI division have transitioned to roles within the generative AI product division at the company, with some joining the AI infrastructure team. https://bitly.ws/32Uxj 

🔧 Cool Tools and Experiments

  • Adobe’s Project Sound Lift: AI-powered audio separation and editing https://tinyurl.com/2jfmt6u6 

  • Eleven Labs releases Speech-to-Speech. Now users can transform their voice into another character with full control over emotions, timing, and delivery.  https://lnkd.in/eg6_ZmcR

  • Runway releases features and updates, including Gen-2 Style Presets that allows users to generate content using curated styles without the need for complicated prompting, Motion Brush, updated camera controls, and improved image model https://tinyurl.com/426cuvxc 

  • Stability AI announces Stable Video Diffusion, an AI model that generates videos by animating existing images https://tcrn.ch/3GbOgVJ 

  • Microsoft releases Azure AI Speech, a new text-to-speech software that that can generate new avatars and voices or replicate the existing appearance and speech of a user https://tinyurl.com/wbs7b9ds 

💰 Latest Startup Fundings

  • CreateSafe, the company behind Grime’s voice cloning tool, raises $4.6M in seed round to introduce a new AI-powered platform that will help artists monetize generative AI opportunities in music creation and distribution https://tinyurl.com/ta227xej