It seems like you only need two things to win the AI arms race: scale and money.
Large, well-funded AI developers are focused on collecting massive amounts of data, developing computing power, and building ever-larger models.
But that's not the only way forward.
“It's absolutely true that the more math you put into a model, the bigger the model, the better the model will be,” Cohere CEO Aidan Gomes said in a recent interview with 20VC. “It's the most reliable way to improve a model, but it's also the dumbest way.”
He said there are several other ways small businesses can compete.
“There's a lot of pressure to create smaller, more efficient and smarter models through data, algorithms and methodologies, not just through market forces to scale,” he said.
Gomez said many of the advances in the open source space have also come about through improvements in how data is processed, including better algorithms to gather higher quality data from the internet and innovations in synthetic data.
Cohere, which develops enterprise AI technology, was valued at $5 billion in its latest funding round in July. Earlier this year, the company unveiled a family of language models called Command R, which COO Martin Kon previously told Business Insider was part of an “emerging category of scalable models.” As the company continues to deliver enterprise technology, Kon said affordability and reliability are more important than providing businesses with the latest AI technology.
