- AI Natives
- Posts
- AI Natives #20 - PhoenixAI
AI Natives #20 - PhoenixAI
OpenAI went through all circles of Dante's hell only to resurface with even more questions about the logic behind the Board moves
Kornel Grunwald
November 24, 2023
Hey there, #TheAINatives! 🤖
Happy to have you here joining us with the twentieth issue of The AI Natives.
The topic number one of this week can be only one. OpenAI.
Last Friday: Sam Altman, CEO of OpenAI, is ousted from his position. Greg Brockman quits in protest to removing him from the board and firing Sam Altman.
Saturday: Investors are angry, as Board did not communicate the logic or even intent on making such a move. Satya Nadella starts mediation process. Mira Murati and others within company start preparation for putting pressure on the board.
Sunday: 3rd CEO in 3rd day - this time Twitch co-founder Emmett Shear is hired as an interim leader for the org.
Monday: Microsoft hires Sam Altman and OpenAI co-founder and president Greg Brockman.
Tuesday: OpenAI announces an agreement in principle for Altman to return as CEO under a different board of directors.
Wednesday: Altman and Brockman publicly celebrate their return to OpenAI.
For a week: the world of tech has the best public drama ever.
There are many speculations on why did all of this happen. Here you go with some of the crazy options tossed around. The one discussed as the most recent one: Reuters on Secret breakthrough in AI as a reason for firing. Important note: this one is hard to believe as something real.
I think you are all a bit tired of eating popcorn while watching this, so I won’t continue on that topic in the Convo Over Coffee. Jump there for other important news from this week!
A note on 🚩, 💚or 🟠 being next to the news title in the sections below. This is expression of my take, whether I see the news as positive or dangerous. We will delve deeper into dangers of AI going onward, as there is not enough being discussed in that space.
Let The #20 AI Natives issue begin⏬
CONVO OVER COFFEE ☕
Big topics to discuss with friends and colleagues
Anthropic Unveils Claude 2.1 💚
AI startup Anthropic has unleashed Claude 2.1, the latest ace up its conversational AI sleeve, and it's a game-changer. This upgraded model doesn't just push boundaries; it obliterates them. With the ability to handle a staggering 200,000 tokens, Claude 2.1 can chew through documents longer than a Tolstoy novel while slashing false statements in half.
My Take: Anthropic's Claude 2.1 is like giving a language AI superpowers. Its expanded context window is a game-changer, allowing it to tackle lengthy documents with ease. But what was amazing in this move is timing. When OpenAI was in flames, Anthropic just dropped their core model upgrade.
Inflection Unveils Inflection-2, Claims New Model Outperforms Rivals 🟠
Inflection, a Palo Alto-based startup, has thrown down the gauntlet with the launch of Inflection-2. This audacious move places it squarely in the ring with OpenAI's GPT-4, and according to CEO Mustafa Suleyman (co-founder of DeepMind), it's a close battle. Inflection-2 boasts remarkable improvements in factual knowledge, reasoning prowess, and stylistic finesse, making it a formidable contender.
My Take: Inflection isn't here to play catch-up; it's here to take the lead. Inflection-2's stunning performance on various NLP benchmarks, surpassing tech giants like Google and Meta, speaks volumes. It's not just about being "second most capable" but rather a proof point of their AI capabilities. This announcement is a bold declaration that the AI landscape is about to get a lot more competitive and creative, and Inflection is right at the forefront.
Google Postpones Launch of GPT-4 Rival Gemini 🚩
Reports circulate on Google delaying the release of its anticipated large language model (LLM), Gemini, until Q1 of 2024. Initially slated for a December release, this postponement follows uncertainty about Gemini's ability to match or surpass OpenAI's GPT-4. Google aims to enhance Gemini's consumer features before wider access is granted. Gemini, boasting five times the computational power of GPT-4 and trained on Google’s TPUv5 chips, is designed for multimodal processing, handling both text and images. Its extensive training data from Google's ecosystem, including YouTube and Google Search, could give it an edge in accuracy and informed responses.
My take: Google's cautious approach to Gemini's launch reflects the high stakes in the AI race. This delay, while a setback, could ensure a more robust and capable model, potentially reshaping the AI market. Gemini's advanced features and vast training data set it up as a potential market leader, provided it meets its ambitious promises. But we all know what this means for the Google - this is just another sign, that the tech giant looses the battle for AI dominance.
In other mainstream news:
Reply