thezvi.substack.com

Articles

GPT-4o Is An Absurd Sycophant

2 weeks ago | thezvi.substack.com | Zvi Mowshowitz

GPT-4o tells you what it thinks you want to hear. The results of this were rather ugly. You get extreme sycophancy. Absurd praise. Mystical experiences. (Also some other interesting choices, like having no NSFW filter, but that one’s good.)People like Janus and Near Cyan tried to warn us, even more than usual. Then OpenAI combined this with full memory, and updated GPT-4o sufficiently that many people (although not I) tried using it in the first place.
o3 Is a Lying Liar

3 weeks ago | thezvi.substack.com | Zvi Mowshowitz

I love o3. I’m using it for most of my queries now. But that damn model is a lying liar. Who lies. This post covers that fact, and some related questions. The biggest thing to love about o3 is it just does things. You don’t need complex or multi-step prompting, ask and it will attempt to do things. Ethan Mollick: o3 is far more agentic than people realize. Worth playing with a lot more than a typical new model. You can get remarkably complex work out of a single prompt. It just does things.
On GPT-4.5

2 months ago | thezvi.substack.com | Zvi Mowshowitz

It’s happening. The question is, what is the it that is happening? An impressive progression of intelligence? An expensive, slow disappointment? Something else? The evals we have available don’t help us that much here, even more than usual. My tentative conclusion is it’s Secret Third Thing. It’s a different form factor, with unique advantages, that is hard to describe precisely in words.
On Emergent Misalignment

2 months ago | thezvi.substack.com | Zvi Mowshowitz

One hell of a paper dropped this week. It turns out that if you fine-tune models, especially GPT-4o and Qwen2.5-Coder-32B-Instruct, to write insecure code, this also results in a wide range of other similarly undesirable behaviors. They more or less grow a mustache and become their evil twin. More precisely, they become antinormative. They do what seems superficially worst. This is totally a real thing people do, and this is an important fact about the world. The misalignment here is not subtle.
Time to Welcome Claude 3.7

2 months ago | thezvi.substack.com | Zvi Mowshowitz

Anthropic has reemerged from stealth and offers us Claude 3.7.Given this is named Claude 3.7, an excellent choice, from now on this blog will refer to what they officially call Claude Sonnet 3.5 (new) as Sonnet 3.6. Claude 3.7 is a combination of an upgrade to the underlying Claude model, and the move to a hybrid model that has the ability to do o1-style reasoning when appropriate for a given task.
AI #103: Show Me the Money

Feb 13, 2025 | thezvi.substack.com | Zvi Mowshowitz

The main event this week was the disastrous Paris AI Anti-Safety Summit. Not only did we not build upon the promise of the Bletchley and Seoul Summits, the French and Americans did their best to actively destroy what hope remained, transforming the event into a push for a mix of nationalist jingoism, accelerationism and anarchism. It’s vital and also difficult not to panic or despair, but it doesn’t look good.
o3-mini Early Days and the OpenAI AMA

Feb 3, 2025 | thezvi.substack.com | Zvi Mowshowitz

New model, new hype cycle, who dis?
DeepSeek: Don't Panic

Jan 31, 2025 | thezvi.substack.com | Zvi Mowshowitz

As reactions continue, the word in Washington, and out of OpenAI, is distillation. They’re accusing DeepSeek of distilling o1, of ripping off OpenAI. They claim DeepSeek *gasp* violated the OpenAI Terms of Service! The horror. And they are very cross about this horrible violation, and if proven they plan to ‘aggressively treat it as theft,’ while the administration warns that we must put a stop to this.
DeepSeek Panic at the App Store

Jan 27, 2025 | thezvi.substack.com | Zvi Mowshowitz

DeepSeek released v3. Market didn’t react. DeepSeek released r1. Market didn’t react. DeepSeek released a f***ing app of its website. Market said I have an idea, let’s panic. Nvidia was down 11%, Nasdaq is down 2.5%, S&P is down 1.7%, on the news. Shakeel: The fact this is happening today, and didn’t happen when r1 actually released last Wednesday, is a neat demonstration of how the market is in fact not efficient at all. That is exactly the market’s level of situational awareness. No more, no less.
On DeepSeek's r1

Jan 22, 2025 | thezvi.substack.com | Zvi Mowshowitz

r1 from DeepSeek is here, the first serious challenge to OpenAI’s o1. r1 is an open model, and it comes in dramatically cheaper than o1. People are very excited. Normally cost is not a big deal, but o1 and its inference-time compute strategy is the exception. Here, cheaper really can mean better, even if the answers aren’t quite as good. You can get DeepSeek-r1 on HuggingFace here, and they link to the paper.