Hacker News Reader

Comment by reissbaker | Hacker News Reader

Wow. This feels like a regulatory regime designed to smother open-source AI, and reduce competition in the American market since small companies are unlikely to be able to comply. The worst part is that doing so is actually against our national interests... China has already essentially caught up to GPT-4V with the release of Qwen-VL-Max yesterday [1]; the cat is out of the bag. Hobbling ourselves now doesn't help.

1: https://twitter.com/_akhaliq/status/1752033872982806718

View on HN

Yeah the genie is out of the bottle. Any regulations a government sets up just puts its economy behind the competition. I think a lot of the US regulations are coming from large tech companies (google) feeling threatened. LLMs used for knowledge and skills transfer threaten search and therefore ads. If you can make it illegal or impractical then you don’t have to compete.

> Any regulations a government sets up just puts its economy behind the competition.

I fully expect us to do this in Europe. Then complain in 10 years when all the big AI tech comes from abroad.

Well let's say congrats to them then, because we will all benefit back from that, especially since they work more in open way as we can download the Qwen-VL-Max model, inspect it and use it offline.

Europe is copying ChatGPT, and all we got is a louzy LLaMA-fork.

Definitely congrats to them. I don't hate on Mistral though; their open-source models are extremely capable for their size and parameter count, and Mistral Medium is second only to GPT-4 on the Chatbot Arena (I'm not including Bard, since Bard gives their chatbot access to the internet, whereas the other models are being interacted with solely on the basis of their innate knowledge; that doesn't seem apples-to-apples, and the underlying Gemini Pro model without access to the internet underperforms GPT-4 and Mistral). And Mistral Medium is much cheaper than GPT-4 — around $8/million tokens, whereas GPT-4-Turbo is $30/million tokens. So I'd say Mistral is doing a decent job.

With this kind of regulatory regime I wonder how much of the next wave of AI companies are going to be started abroad... I sort of doubt Mistral is the last one, even though the U.S. has such high talent density.

> and Mistral Medium is second only to GPT-4 on the Chatbot Arena

..Chatbot Arena is an incredibly flawed metric. Incredibly biased towards English, and towards certain kinds of tasks people are more likely to test on Chatbot Arena.

If you use a broad range of tasks, let alone in a diverse range of languages, unfortunately GPT-4 is still miles ahead of Mistral. I wish it wasn't the case, hope Mistral catches up as soon as possible.

Would've been great if anyone disagreeing could've highlighted why, but alas. If you've tried Mistral on some creative tasks in a wide range of languages and found it competitive with GPT-4, please do tell.

I agree with you it's great that Mistral exists and the efforts they do are very useful.

I'm sure they are going to improve, but compared to what you have shown it's still more of a garage-shop.

(Is Mistral worth 2B+ EUR ? That's another question).

In one way or another, such regulation already implicitly exist in all the countries where there is no First Amendment equivalent.

Perhaps a workaround would be to claim that this is free speech protected by the Fist Amendment and shouldn't be controlled.

(Back to Mistral and regulations), in France, for example, there are certain words or things you cannot write or say without risking to go to jail (in theory, but in practice it's mostly fines), or if you spread certain historical "fake news".

The whole drumbeat from OpenAI has always looked like using regulation to create a moat out of thin air.

Are any of the open models in danger of disappearing soon? Should I be downloading local copies before it is to late?

I feel like this one is impossible to do right. On one side, it's people like you saying: 'government oversight is hurting our innovation'. Directly opposed is: 'AI Companies are growing too fast and something bad will happen if oversight doesn't come soon.'

I agree with both and really don't have a clue what's the wise thing to do. I think it's probably the oversight thing, but since other countries will definitely compete without this oversight, someone will eventually release SkyNet. But do we want to be the first?

I don't know...

Since there is no regulatory response that can deal with unknown unknowns like your second point, the correct answer is to not regulate for now.

We know that excessive premature regulation by captive government agencies hurts innovation and empowers competing nations. We don't know that AI companies are growing too fast, or that Something Bad Will Happen if oversight doesn't come soon.

Seems like an easy choice. We'd never have left the caves if we allowed the precautionary principle to dictate our lives.

Why are we in a rush to "compete with China" on this front? I don't exactly view that country as the bastions of a government who wants to promote the welfare of its citizens. They already I famously don't comply with copyright but no one is. Clamoring to take out the patent office to "compete".

What is being a few years behind on China going to do to the western economy to make it not worth slowing down and considering the ethical quandaries? We know from last decade how hard it can be to regulate back issues once companies overreach.

If AI turns out to be no big deal, then you're right, there's no rush to compete. And there's also no rush to regulate. Conversely, if AI turns out to to be a huge deal, then we certainly won't end up wishing we'd held ourselves back with regulation while rival/adversary nations plowed forward at full speed.

Look at the people who will be doing the regulating. They'll get right on it after they decide which flavor of genocide to support, what books your local library gets to purchase, and who gets to use what restroom.

Basically there is almost zero upside to early government intervention, but the potential downside to regulation is unbounded. That makes it a very easy call, to my way of thinking.

Didn't China also issue a ruling that AI should conform to proper socialist values? I'm not sure how they are enforcing that though, Biden's executive order seems to have a lot more teeth (and even then, there is no standard yet, they just want information shared).

I can appreciate that the model might be trained to work on primarily chinese language, but the performance for me really doesnt seem to be there yet.

I hate to do the "I asked ai and pasted it here" style of comment, but I really wanted to understand the validity of the claim it surpasses gpt4 in all those benchmarks. I asked gpt4 and qwen-vl-max this prompt to try to get a sense of their capabilities. Nothing even close imo.

"tell me a story about relations between people living on one side of a penny floating through space and the other. Use the format of the hero's journey, including rejecting the call, focus on worldbuilding by hinting at further complexity, and do it in a kafkaesque style"

quen's story starts with: Once upon a time, there was a penny floating through space.

gpt 4: Through the vast emptiness of space drifted an ancient penny, a relic from a long-forgotten Earth.

The difference is night and day atm. Ask yourself: would Kafka have started a story with "once upon a time"

No, because all of Kafka's books were written in German.

I don't think we can extrapolate much from that anecdote, but I'll try. The GPT version was predictably wordy and had the signature precocious high-school essay feeling, e.g. lots of adjectives. It didn't feel Kafkaesque at all, where is the confusion, breathless tone, mysterious K.?

Qwen's is pretty unremarkable, just very direct, but I don't see that one is actually following directions better. Maybe further along the difference gets more stark. Also, isn't Qwen a Chinese-first model? How much of the fine tuning was in English?

Neither example gets too actual utility, though it could highlight the sort of demo that chatgpt is trained for.

that's fair. I just didn't want to paste a massive text block so went with the first line, which showed the extremely basic way of starting quen used. Link to the pastebin below (actually ended up being anonpate, seems like pastebin is down for me?)

I felt the overall writing & characterization of gpt4 was far better, including themes of bureaucracy and psychosis whereas quen sounds like a cliched 6 year old writing an aimless story.

I'm not saying either of these are amazing stories, but I think any english teacher/creative writing teacher would rate one over the other as things stand right now. Things like the comparing the rows of wheat embossed on the penny to the regimented way of life its people lead are actually half decent.

https://anonpaste.pw/v/ccbfb3e8-746f-4311-9db4-cedbf6c3a440#...

Reply on news.ycombinator.com