Yes, but considering the goal of his project is a proof of concept, that is, proving this project is possible, acid3 is a good place to start to prove the worth of the project and aiming 100% first make more sense. Tweaks will be needed one way or another to make it equivalent to the big 3.

For a piece of software designed by reading the specs and implementing accordingly, there acid3 would be a good way of demonstrating the quality of the result.

But is that what happened? If you pass tests that conflict with the specs, then that implies that the implementation is not based on specs, but on feedback from the test. Which means it may be designed to pass the test, and not much good once you go outside the test examples.

I'll extend the benefit of the doubt and say that it's possible that the implementation is based on an older version of specs, one that matches the tests. But it doesn't inspire confidence that all we're shown is a single screenshot and no source code.

I'm somewhat skeptical of that approach. It's not even clear if "world models" are significant for our subcouscious decision making as human beings. Why assume it's necessary for AI?

Maybe it is assumed in this video, but Yann Lecun and many other demonstrated how physics remains a big, completely unsolved challenge for LLMs with no solution in sight due to its nature.

But I'm not at all an expert in the domain and I might have misunderstood something please feel free to contradict me!

How would understanding the environment of a decision not matter for making a good decision?

Let's suppose each models was subsidized at 70%, so that we only pay 30% of the cost. They would loose much more money per token on the more powerful models. It's in their interest to encourage the use of the less expensive models. Let's say they increase Luna subsidies at 90%. They would still "save" relative to the use of the more expensive models.

> Let's suppose each models was subsidized at 70%, so that we only pay 30% of the cost.

why on earth would you suppose that?

Because their leaked financials point that way.

Source?

The company losing money does not mean the model inference in API is 70% subsidized- that’s a crazy leap in logic. Obviously the massive number of free chatgpt users are getting subsidized 100%.

Not true at all. Do you really believe that generating 1m output tokens on sol really cost 100 usd?

These companies are printing money on inference. The only issue is the capex of expending on new servers but they are generating TONS of income.

Any locks can be picked given enough time and it might be the situation we are in with IT security. I'm surprised it's not an already common practice of spreading terabytes of fake data, fake keys, and fake servers and so forth. Slowing down AI attacks will become important. Monitoring access to fake data and triggering kill switch should be an no-brainer. Obsuscating libraries and tools names is another one.

You might like https://mirage.io/blog/bitcoin-pinata-results

As far security: you can get a lock further, if you are willing to prove your code safe and secure. Thanks to LLMs that no longer requires a PhD.

I don't mind it as long as the company engage in the comments. They haven't so far.

here now.

From Google point of view, their users are nothing more than a statistic. An hacked account is a statistic. No surprise they see software that way.

Greed and inefficiency is what define SaaS. Compare what Microsoft give you for 50 bucks per month vs Atlassian or any ticketing services. It's absurd. They can charge up to $80/month for what shouldn't cost more than $5.

Looking at some of the pricing and then multiplying it by 12 and ask question would you pay this for shrink wrapped software is bit eye opening. At least from not filthy rich Europen perspective.

Not to say that some products like MS aren't seemingly very good value. But on other side like chat and source control I do raise eye brow...

In Quebec, new data center will have to pay almost twice the price of what citizen are paying. it's 14 cents/kwh vs about 8 cents/kwh. Crypto center are set to pay 18 cents/kwh. That's the advantage of having a public utility.

They add a mode to Codex which is less technical or more oriented toward office work and you can toggle between the two.

Literally no UI changes when I toggle between the two. It's 100% identical, one is not less technical than the other.

The button changes from green to blue.

Is it a dumb-down version of GPT like the current voice model? At least in french, I find the current GPT voice mode to be useless, to the point I only use the dication mode. I would ask a question and it would answer something along "That's a interesting question. I can help you with that. Anything you want to know about X?" I would ask again and it would answer the same kind of non answer.

Click through to the link, the answer is no it uses the latest gpt models now.

> the answer is no it uses the latest gpt models now.

Actually it says it _can_ delegate to the latest models. Seems reasonable to ask how the voice model does when it doesn't delegate (or while waiting for the delegated answer).

I haven't stress-tested it, but I would imagine it approaches complex problems the same way a human with a phone in their pocket would — that is, by having a degree of awareness of the confidence it has in its own knowledge in some areas; where, when it "realizes that it doesn't know", it blocks the conversation with statements like "I don't know, let me check."

I say this because this is already how ChatGPT works internally when using its "auto" mode; the version of the "fast" model used in the "auto" mode does the same "notice your ignorance and bring in the heavy model" thing, just silently, rather than mentioning that it's doing it.

(If someone has actually run the experiment, please chime in!)