Aye! Meta tried recruiting in the ML space. The hook? Half of the interviewers were Microsoft alumni that had a ton of money thrown at them. I wonder where those folks are today.
MS bet on OpenAI is one of the biggest tech success stories of this decade for sure. Now everyone wants a pie of it but it seems that they are way behind for some reason (for example Google Bard is nothing compared to OpenAI implementation)
Yes, it's quite ironic isn't it that the big AI progress is being made not by the cool kids at Alphabet and Meta and the like, but in a place backed by "boring" old Microsoft. Whatever next - Bing becoming actually useful, perhaps?
They did; perhaps not in ways that were immediately obvious to the general public. Back in 2014, they created attention to improve how RNNs handled long sequences, an effort driven by google translate. Same goes for when they created the transformer architecture. The first product to see an improvement was also translate. They simply never exposed the bare language models to the public a la OpenAI style with GPT3
While GitHub copilot is a big deal, it is a tiny tiny tiny portion of the promise of LLMs. There will be far more economic disruption from LLMs as productivity tools than as coding tools.
> Google Bard is nothing compared to OpenAI implementation
That's simply not true based on the comparison done by MKBHD [1].
Bard has the requirement that it needs to deliver an accurate and quality result above everything else. This makes it more useful but far less interesting and compelling than OpenAI.
It's true based on its demonstrable lack of ability to solve tasks that even ChatGPT (3.5) can handle. Sure, you can cherry-pick some test suite to come up with a number that says otherwise - just like there are people claiming that LLaMA can match ChatGPT - but this is more of a "squinting at it right" kind of thing.
I don't know of anyone using Bard over ChatGPT in any application where quality of the output matters.
