The biggest question facing artificial intelligence: Is it more like Google or a bag of rice?

• According to Marc Andreessen, there are two competing visions for the future of the artificial intelligence industry.
• First, it is an extremely profitable, winner-takes-all market. Second: It’s a race to the bottom.
• OpenAI’s chief product officer isn’t worried that open source development will cause the latter situation.

Is artificial intelligence more like Google or a bag of rice?

This is not a puzzle. Rather, this is one of the most fundamental (and arguably surreal) questions facing the AI ​​industry today, according to legendary investor Marc Andreessen.

Speaking this week at the Ray Summit, an AI conference in San Francisco hosted by Anyscale, Andreessen said there are two competing visions for the future of the AI ​​industry.

In the first vision, AI startups are engaged in a winner-takes-all race. This will lead to “monopoly and infinite profits due to scale. The biggest company will have the best model that answers all the questions better, and then they will charge whatever they want for it,” Andreessen explained. “That’s what happened with search, where Google ended up in this position.”

The future of alternative AI is not so bright for profit-hungry AI entrepreneurs. Hypercapitalized startups “can be in a race to the bottom where you actually find that selling intelligence is selling rice or something,” Andreessen said, with slim profit margins and little product differentiation.

“It turns out that anyone can create an LLM, there are open source LLM projects, new LLM startups appear every day and it turns out that anyone can search the Internet, anyone can buy GPUs and then basically anyone can get the same result,” he explained, referring to large language models, which are a common type of AI model.

Related stories

Rice or Google?

Is OpenAI creating the next Google search engine? Or is he spending billions of dollars to get into the rice selling business?

We received the answer to the question of what OpenAI thought about it the next day at the Ray Summit. This was on the same day that the startup closed a $6.6 billion funding round at a post-money valuation of $157 billion.

On Wednesday, OpenAI Chief Product Officer Kevin Weil took the stage, where Anyscale co-founder Robert Nishihara asked him what open source models mean to OpenAI’s business.

These AI models, such as Meta’s Lamy offering, are freely available for almost anyone to use, while OpenAI provides access to relevant, closed models. The problem is that cheaper alternatives are good enough for most users, which could undermine OpenAI’s ability to charge for access to its best GPT models. Indeed, the price that AI companies can charge users for a “token” has dropped over the past 18 months.

Weil said he wasn’t concerned about it.

“I’m glad open source models exist. I mean, I don’t think it’s an either-or rule. From a philosophical mission standpoint, it’s about getting more AI into the hands of more people, and I think that’s great,” he said, referring to Meta’s approach in his interview with Lama. “I worked closely with Mark (Zuckerberg) for many years. I have great respect for him. This is a super smart strategy for Meta.”

He added that OpenAI is making some open-source efforts, pointing to the Whisper audio transcription model.

Weil highlighted OpenAI’s superior technology, arguing that “ultimately, people will want to find the most efficient models at the best price, while also being as secure as possible.” Later, discussing the start-up’s new o1 model, he added: “Before people catch up, we will try to be three steps ahead.”

Production cost

But at the end of the on-stage interview, Weil acknowledged that AI companies are limited in their ability to charge high prices for their best models – no matter how useful the final products are.

“The world is going to change, and if it becomes possible to do these things for $3, you won’t be able to charge $5,000 for them for a long time because someone will come in and charge you very quickly,” he said. “These things end up around the actual cost of production.”

OpenAI is growing rapidly, but it’s still losing a lot of money. With nearly $7 billion in the bank from outside investors, there may soon be more pressure to turn a profit or at least make significant progress toward that goal.

However, if downloadable open source models bring all prices down to “cost of production,” then Andreessen’s first vision of “infinite profits” will start to seem a little less likely.

Instead, it’s starting to look like selling rice.

Do you work at OpenAI? Contact this reporter via the Signal encrypted messaging app at +1 650-636-6268 Or write to rprice@businessinsider. We can keep sources anonymous. To contact us, use a non-work device.