HN Explorer

Debunking the Microsoft/OpenAI/Google hype and over-reaction

18 points by eclectic29 ↗ HN
Satya got lucky with OpenAI investment. Nothing smart. Now he’s acting like he foresaw OpenAI's potential and invested in the right place. I mean I’m happy the stock is rising but I know this is too early to say Microsoft won the AI wars and Google lost. Media is stupid and so are many techies who are overreacting. In fact even saying lucky is a bit too premature at this point.

Obviously Google botched the demo and should’ve been a little proactive but it’s still no grounds to believe that it’s behind Microsoft in AI. Having said that it still doesn’t take away the fact that Satya is much more aggressive than the coasting Sundar. Sundar truly deserves to be demoted if not fired.

Frankly we’re just like the ops division of OpenAI. We let them do research on our Azure clusters. Although I work at Microsoft I like to say it what it is.

It’s a bit alarming for Satya to come out and say they made Google dance. People forget that it’s was the seminal transformer paper released by Google in 2017 that changed the NLP industry dramatically. It’s the core backbone of ChatGPT.

I must say that I’m really happy to read all the positive sentiments about Microsoft which until yesterday was the football of even the shittiest companies out there.

10 comments

[ 2.8 ms ] story [ 34.7 ms ] thread
loveparade
Unfortunately that's how tech and media works. OpenAI itself was lucky as well. Before the LLM success, they were just another research lab that did the same stuff as everyone else and were mostly focused on RL. A few researchers randomly stumbled upon [0] the fact huge LLMs work unexpectedly well, but this might as well have happened at any other lab like Google or Meta because everyone was doing the same kind of research anyway. Anyone who is acting as if any of this was planned is just spinning a narrative.

[0] https://news.ycombinator.com/item?id=34754355

p1esk
they were just another research lab

They have been a top-3 AI lab right from the start.

A few researchers randomly stumbled upon [0] the fact huge LLMs work unexpectedly well

No. Early on Sutskever said he believed they can get to AGI by simply scaling up models. Yes, everyone was doing similar research but OpenAI has been consistently first to show groundbreaking results. GPT 1-2-3, DALLe 1-2, CLIP, Jukebox - one of these could be considered luck, but all together points to greatness.

rvz
I would say it is a smart move that gets them slightly competitive with Google since they bought DeepMind. Microsoft was so overlooked in 2011, that it was outside of FAANG. It just means that Microsoft has gotten smarter under Satya's leadership by offering services to developers and owning developer platforms to enable that and this seems to be paying off well.

OpenAI is essentially a Microsoft AI division and having a AI model behind a SaaS is hardly revolutionary. I expect this AI hype to be short lived since both Microsoft and Google still have their AI's hallucinating inaccurate information; making them untrustworthy, unreliable and they are admittedly still experimental.

> Media is stupid and so are many techies who are overreacting. In fact even saying lucky is a bit too premature at this point.

Yes, the media does this all the time to generate FUD and nonsense, and they did this to Microsoft before. Thus, it is an absolute over-reaction to immediately declare that Microsoft has 'won' with OpenAI. It just means that this time, Google finally has a serious challenger. This rivalry with overtaking Google dates back since Bing and the 'Scroogled' campaigns for Outlook vs Gmail ads.

In reality, it would take more than just OpenAI to dethrone Google completely. ChatGPT itself won't totally replace search engines. In fact Google would be in a worse position if ChatGPT or the GPT-4 model was already open sourced.

The endgame is an open source AI model that is better than ChatGPT but small enough to fit on a smartphone.

bruce511
I get the urge for people to dump on Google. Search makes good profits and makes a tempting target.

And ChatGPT is cool in a "look my computer can write" kind of way. But it's kinda tangential to search, is (currently) laughingly inaccurate, and doesn't seem like an ideal ad platform (yet).

Which leaves me wondering, is this the next "touchscreen" moment, or is it the next "3D TV" moment?

If course it'll find usefulness going forward. It'll get better, and hopefully more accurate. But with my jaundiced eye, trained on endless hype-and-fail cycles, I've yet to be convinced.

One thing's for sure though, it's certainly captured the media's attention. Seldom have I seen a tech get into the public head as fast as this.

It writes very well, and the layperson does not expect their computer to lie, so if you accept what it says is true, then it really looks amazing.

Silverback_VII
Google has with LaMDA a similar or even better system than ChatGPT but obviously, they are losing the race because they are too busy making it as political correct and harmless as possible.
muzani
I think it's a steam engine moment, similar to the one by James Watt. ChatGPT is something actually commercially viable after decades of research on AI.

Replacing search would be like replacing horses. The first locomotive was about 2 decades after the Watt engine. But the Ford assembly lines were what really replaces horses, about 137 years later.

Of course technological progress today is a lot faster. But probably slower than everyone expects. It's not likely to kill Google next year, but PageRank style searching will likely be replaced with conversational style questioning given enough decades.

rl3
>Having said that it still doesn’t take away the fact that Satya is much more aggressive than the coasting Sundar. Sundar truly deserves to be demoted if not fired.

https://news.ycombinator.com/item?id=34716375

If that's true, I'd say it's the other way around.

Granted, in my opinion the world would be a better place if both companies ceased to exist.

encryptluks2
Agreed, but Microsoft is the type to use AI to run manipulation campaigns. I remember them using NVIDIA Teslas for AI back in 2015 and loading Reddit big data to manipulate social media and counter negative news about them.
seydor
But BingChat remains THE super-app. Given the right resources, it can potentially find me the best flight combo for my next trip in a single go. I don't want a list, just the final result and approve or not. It doesnt really substitute search, it replaces it with finding the perfect answer.

I assume MS/Google/OpenAI will launch an app platform on top of this. We will contribute some form of data and get paid back according to some formula, and this is what will replace google, and largely the web as we know. I dont know if MS has announced something like that but seems like the obvious way forward, and it seems like a big business.

I think the hype is more about the potential of this tech rather than the MS/google pissing contest which is typical media entertainment