Key industry figures have already spoken at YC AI Startup School: the CEO of OpenAI, the co-founder of Anthropic, and the head of Perplexity. And they have very different opinions!
Sam: Our models are the best on the market; they will become even more outstanding. Invest everything in startups built on our platform!
Kaplan (Anthropic): Calm down about AGI. Don’t forget about safety—safety—safety. It’s better if we handle the training of AGI ourselves, it’s safer and more secure.
Shelle: We’re still a long way from true AGI. Here’s a comparison explaining why that is. And here’s a benchmark showing the reasons.
Of course, there were many motivational talks and discussions about promising future developments—without the technical jargon typical of events like CVPR.
But the most interesting presentation was by François Shellé—he discussed the weaknesses of current models like GPT-3 and what they lack. I’ve long wanted to write a separate post about this.
By the way, I feel like there are at least 10 readers here who know me personally. Let’s organize a small meetup in San Francisco—join the chat: https://t.me/+E2Elz7FxXTdlMTMy