The advertising cash firehose is perhaps the only thing that matters at Google.
But, having been severely caned by ChatGPT and left flat footed on AI i think it’s reasonable to say even that firehose is now at risk.
Execution is more important than being the first. MS and OpenAI did an excellent move, and now Google needs catch up. It doesn’t matter if they invented the model architecture used by ChatGPT, they are not the ones capitalizing it.
Case in point, 2017 Google Brain released the "Attention Is All You Need" paper however haven't done anything proactively to deliver it to mass users like OpenAI have done (among others) - perhaps this is the kick that Google needs to remind them that research alone isn't enough, it needs practical applications and the question could be whether Pichai is the right CEO to deliver that
etc.
history is littered with the dominant player getting shellacked by an upstart
And it usually is for political reasons not technical reasons.
Bard is a piece of shit. It's probably being held back so that it doesn't compete internally with google search
This pattern plays out all the time in the tech industry. You think google didn't know how to make a website like facebook? Of course, but they couldn't beat a much smaller weaker opponent even then. Google now is about 8 times larger than Google 2011 and much slower.
Unfortunately, this recent LLM trend will seriously threaten it. If an LLM can provide the answer, then why would a user click to go to a website? And if there are no clicks to a website, why would the website curate the information? How will they make money? The long-term effects of how LLMs are integrated into Search are still unknown. Of course, MSFT would love to push Google into LLMs, as that hurts their ads business.
Also (as I mentioned in an earlier comment) he has been milking that cow a bit too much. The ad load on Google sites has gone through the roof. A fair(er) comparison of his performance would be to see how much revenue Google would make at the same ad load as in 2015. Any idiot can make 5x more revenue in the very short term by showing 20x more ads. That is not rocket science. What _is_ interesting is if you can do it without increasing the ad load. That's where the real magic lies.
