80% of their revenue comes from ads, which is google search. Chatgpt and any other startup who invests in something like chatgpt can upend googles core business model.

and no, google doesn’t have anything significant in this space. “Innovation” behind search was to slowly increase no. if paid ads in your results, driven by need to increase revenue and keep the rest of the 80% of biz afloat.

> and no, google doesn’t have anything significant in this space

LAMBDA and PaLM are literally the answer to Chat-GPT. Google is their closest competitor.

I do agree that they will lose their moat, and soon Google search will just be another tool among many.

And Google developed LaMDA model (superior to chatGPT) a lot of time before OpenAI, additionally without their research into transformers and publishing results, OpenAI would probably not even have ChatGPT now/yet.

Something I find very curious - Google are doing the same kind of research and are in many ways futher along. They have thier own models like T5 that supposedly perform better. Are their exec unaware of their own strengths?

For that matter, wasn't Google collecting masses of GIS data and doing intense self driving car research? Where is that product?

Serious q - do google have so much money that they are doing tons of research and then just forgetting to make a product out of it?

> For that matter, wasn't Google collecting masses of GIS data and doing intense self driving car research? Where is that product?

https://waymo.com/

Research does not need to lead to a product to be profitable. Another strategy is to patent all of your work, then steal the profits from anyone that attempts to enter that market.

> Why are they in a panic over ChatGpt?

The NYTimes and the rest of the media misread between the lines on this story.

Google aren't worried about competition. They are worried about losing the war with SEO.

ChatGPT showed them a glimpse of the SEO endgame. It upsets the balance between search ranking/boosting and SEO. Whatever adaptations they put into their boosting algo, GPT-style AI can eventually learn. Once learned, it can be subverted.

Fundamentally search boosting is at a disadvantage, because SEO has (blackbox) access to the boosting algo by feeding content to the crawlers and running searches. However the booster doesn't have access to the SEO algo; in fact it doesn't even know (beyond a guess) how many discrete SEO algos it is fighting, or which sites are optimized by which algos. So there is great asymmetry there.

That is why the fire alarm was pulled.