Tweets

2024-10-16T15:37:14+00:00 | 🔗

For anyone who thought that Yann and Sam were apart on timelines, I would recommend outsourcing your "reasoning" to chatgpt. Which is longer two thousand days or 5 years? https://t.co/mjilY15b6f https://t.co/OL65786pN4

2024-10-16T14:47:24+00:00 | 🔗

If a tree falls in a forest and no one is around to hear it, does it make a sound? https://t.co/ZbZSNtnKjG

2024-10-16T14:26:46+00:00 | 🔗

I want to ask if anyone has actually *read* the work these PhD-level LLMs are producing (autonomously).

2024-10-16T14:25:07+00:00 | 🔗

Where do you sit on the "it depends" to "conviction" scale? https://t.co/gGMKIijV6S

2024-10-16T14:14:58+00:00 | 🔗

The other way to look at this is that IF new iterations of foundation models, e.g. GPT-5, were going to be soooooooo much better, we wouldn't have so much clutter in the 4 variants. https://t.co/uLObSBmiKt

2024-10-16T14:08:29+00:00 | 🔗

Now significantly biased against autocomplete.

2024-10-16T14:08:29+00:00 | 🔗

I stopped using autocomplete for code and other people will too. As products become better at one-shotting applets with infrastructure attached (Replit Modal Artifacts Terraform) we will touch code way less. Simultaneously the domains where we are still touching code are… https://t.co/j39Fezv0ni

2024-10-16T13:59:03+00:00 | 🔗

Collectively our job on Twitter is to accelerate Musk’s cycle time to realize his political opinions are idiotic. The core idiotic opinion being that he trusts Trump and thinks that he has lasting favor. Elon is fundamentally a subordinate of Trump, not a peer like Putin or Kim. https://t.co/v1rvXOlXic

2024-10-16T13:50:58+00:00 | 🔗

It’s fascinating to see the “will the LLM make my company look like an idiot and maybe get cancelled” tests rebranded into safety tests for the good of humanity.

2024-10-16T02:30:36+00:00 | 🔗

RT @JustinWolfers: If we're about to fight an election on inflation, it's worth noting that a robust majority of Wall Street economists exp…