Hacker News Reader

Comment by chvid | Hacker News Reader

Not much of a monopoly when they are being outcompeted by open source models done by companies with a fraction of OpenAIs resources.

Sure - a government sanctioned monopoly might have been their fantasy but now it is getting obvious it is not the way things are going to go …

View on HN

> by companies with a fraction of OpenAIs resources.

Meta and Google are orders of magnitude greater than OpenAI and they're the only contenders that have challenged GPT models.

> But what about Mistral et al?

They are not even near GPT-4o. Good gpt-3.5-turbo alt tho.

You might be interested in https://news.ycombinator.com/item?id=42514633

In fairness, OpenAI has been leading benchmarks for the most part, but open source is never more than 3-6 months behind.

Just because some one has a slightly better product doesn’t mean they have a monopoly. That’s like saying McDonalds has a monopoly because they have better fries. It’s just not true in any part of the statement

They have a monopoly on the name McDonalds.

If McDonalds had a fry machine and lobbied that making fries anyone other way was dangerous and only they should be allowed to use the fry machine that would more align with this situation.

It's not a monopoly but trying hard to get a government based one.

You're stretching so far to attempt to make a point you're now trying to equate a trade mark as a monopoly. These are totally different concepts and you've totally lost the plot with this analogy

McDonald's is considered an oligopoly, where a few firms dominate an industry and can set prices. McDonald's is not a monopoly because it doesn't sell a single unique good.

But not even near X SOTA model doesn't matter to 99% of people because 99% of people don't need or care about the difference.

It's like sure Porsche / Ferrari makes a car that can hit insane top speeds really fast on a track, but 99% of people don't drive in such a manner where that matters. In that analogy though, everyday cars aren't free, so profit possibility is even less in OAI case than the supercar case

Claude by Anthropic outperforms OpenAI and Meta's models on several benchmarks and for the coding I do, and Anthropic is only 500-1000 people according to Crunch base

And it's less likely after the election. Trump pledged to repeal Biden's 2023 executive order on AI. He put out a very different executive order in 2019 that sought to reduce barriers. He is an enthusiastic deregulator compared to the current chief. That helped more than a little to get Andreessen, Musk, etc., to climb on the bandwagon. It can all be reversed at the wave of a very transactional presidential hand, but the sounds coming from the incoming administration are more favorable to open AI than to OpenAI.

  https://www.rstreet.org/commentary/ai-policy-in-the-trump-administration-and-congress-after-the-2024-elections/
  https://trumpwhitehouse.archives.gov/presidential-actions/executive-order-maintaining-american-leadership-artificial-intelligence/

Do keep in mind that both Trump and Andreessen & co are not interested in deregulation for ideological reasons, only because this specific deregulation is favourable for the men in question.

It's not about "open AI" but just opposition to the tech giants. And as a consequence, there's a danger that they may pivot on AI regulation the moment anything happens to the current AI bubble that makes it unappealing for VC funding. (OpenAI & co raising the prices on the models to above operating costs would almost certainly eviscerate the AI startup scene)

Reply on news.ycombinator.com