ChatGPT, Google, and the problem of AI’s business model | WARC | The Feed

ChatGPT, Google, and the problem of AI’s business model

ChatGPT is causing some headaches at search giant Google, as similar technology years in the making appears to have been overtaken – but cost issues threaten to return the advantage Google’s way.

Why it matters

ChatGPT is the talk of the internet, one of those tools that feels a bit like magic but it is in the words of OpenAI CEO “eye-watering” in its costs, which threaten to constrain the technology. Google, could have a significant advantage.

What’s going on

Alex Kantrowitz posits on his newsletter Big Technology that Google has treaded conservatively for a company that believes its future and the future of search lies in AI.

• Arguably, Google’s LaMDA is a more potent chatbot – having made headlines when an engineer told the world that it had become sentient – but it also has a much bigger reputation to protect.
• ChatGPT’s inaccuracy isn’t a one off; chatbots often get things wrong, and Google’s entire reputation could suffer heavily if it released to the public an AI that confidently tells people wrong answers.

A bigger issue

How to fund AI bots is now emerging as an issue, one that OpenAI has stated publicly, given the huge costs of the data centres and power needed to run such a system. Roughly, each 750 word tranche of text costs a couple of cents to produce.

However, first mover advantage is powerful, as Kantrowitz explores, and the possibility of ceding the next phase of search before finding a business model could be a risk.

Still, brands and trust matter. Should Google decide to release its system in search (or on its dominant Chrome browser) its sheer availability will be hard to contend with.