Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Google “We have no moat, and neither does OpenAI”

semianalysis.com

Post reply on HN
Print view View original

591–600 of 1001 posts

Re: Google “We have no moat, and neither does OpenAI”

#591

Post by safety1st »

dahwolf wrote: post #389

The current paradigm is that AI is a destination. A product you go to and interact with. That's not at all how the masses are going to interact with AI in the near future. It's going to be seamlessly integrated into every-day software. In Office/Google docs, at the operating system level (Android), in your graphics editor (Adobe), on major web platforms: search, image search, Youtube, the like. Since Google and other…

There are no guarantees about who will or won't own the future, just the observation that disruptive technology makes everyone's fate more volatile. Big tech companies like Google have a lot of in-built advantages, but they're notoriously bad at executing on pivots which fundamentally alter or commoditize their core business. If that wasn't true we'd all be using Microsoft phones (or heck, IBM PCs AND phones).

In Google's case they are still really focused on search whereas LLMs arguably move the focus to answers. I don't use an LLM to search for stuff, it just gives me an answer. Whether this is a huge shift for how Google's business works and whether they will be able to execute it quickly and effectively remains to be seen.

Bill Gates' "Internet Tidal Wave" memo from 1995 is a great piece of relevant historical reading. You can see that he was amazingly prescient about the potential of the Internet at a time when barely anyone was using it. Despite Microsoft having more resources than anyone, totally understanding what a big deal the Internet was going to be, and even coming out of the gate pretty strong by dominating the browser market, they lost a lot of relevancy in the long run because their business was just too tied up in the idea of a box sitting on a desktop in an office as the center of value. (When Windows was dethroned as the company's center of gravity and they put Satya and DevDiv with its Azure offerings in charge, things started to turn around!)

[1] https://lettersofnote.com/2011/07/22/the-internet-tidal-wave...

Re: Google “We have no moat, and neither does OpenAI”

#592

Post by SanderNL »

reissbaker wrote: post #368
tikkun wrote: post #81

The part of the post that resonates for me is that working with the open source community may allow a model to improve faster. And, whichever model improves faster, will win - if it can continue that pace of improvement. The author talks about Koala but notes that ChatGPT is better. GPT-4 is then significantly better than GPT-3.5. If you've used all the models and can afford to spend money, you'd be insane to not use…

GPT-4 is so much better for complex tasks that I wouldn't use anything else. Trying to get 3.5 to do anything complicated is like pulling teeth, and using something worse than 3.5... Oof. TBH this feels like cope from Google; Bard is embarrassingly bad and they expected to be able to compete with OpenAI. In my experience, despite their graph in the article that puts them ahead of Vicuna-13B, they're actually behind..…

SD and its configurability is miles and miles ahead of MJ. Sure if you want a fancy picture now it’s OK. How are you going to generate that same picture in another pose? Inpainting, outpainting.. I don’t even know where to begin. MJ is a toy compared to SD’s ecosystem.

Re: Google “We have no moat, and neither does OpenAI”

#593

Post by reissbaker »

dahwolf wrote: post #389

The current paradigm is that AI is a destination. A product you go to and interact with. That's not at all how the masses are going to interact with AI in the near future. It's going to be seamlessly integrated into every-day software. In Office/Google docs, at the operating system level (Android), in your graphics editor (Adobe), on major web platforms: search, image search, Youtube, the like. Since Google and other…

I think Satya Nadella put it pretty well in an interview: ad revenue, especially from search, is incremental to Microsoft; to Google, it's everything. So while Microsoft is willing to have worse margins on search ads in order to win marketshare from Google, Google has to defend all of their margins — or else they become significantly less profitable in their core business. LLMs cost a lot more than traditional search, and Google can't just drop-in replace its existing product lines with LLMs: that hikes their bottom line, literally. Microsoft is willing to swap out the existing Bing with the "new Bing" based on OpenAI's technology, because they make very little money comparatively on search, and winning marketshare will more than make up for having smaller margins on that marketshare. Google is, IMO, in between a rock and a hard place on this one: either they dramatically increase their cost of revenue to defend marketshare, or they risk losing marketshare to Microsoft in their core business.

Meanwhile, OpenAI gets paid by MS. Not that MS minds! They own a 49% stake in OpenAI, so what's good for OpenAI is what's good for MS.

If Google had decades to figure it out, I think your analysis might be right — although I'm not certain that it is, since I'm not certain that the calculus of "free product, for ad revenue" makes as much sense when the products are much more expensive to run than they were previously. But even if it's correct in the long run, if Google starts slipping now it turns into a death spiral: their share prices slip, meaning the cost of compensation for key employees goes up, meaning they lose critical people (or cut even further into their bottom line, hurting their shares more, until they're forced to make staffing cuts), and they fall even further behind. Just as Google once ate Yahoo! via PageRank, it could get eaten by a disruptive technology like LLMs in the future.

Re: Google “We have no moat, and neither does OpenAI”

#594

Post by weinzierl »

dahwolf wrote: post #389

The current paradigm is that AI is a destination. A product you go to and interact with. That's not at all how the masses are going to interact with AI in the near future. It's going to be seamlessly integrated into every-day software. In Office/Google docs, at the operating system level (Android), in your graphics editor (Adobe), on major web platforms: search, image search, Youtube, the like. Since Google and other…

As running the models seems to be relatively cheap but making them is not I believe that's where the money is. That and generic cloud services because ultimately the majority will train and run their models in the cloud.

So, I would bet on AWS before OpenAI and I would bet the times of freely available high quality models will come to an end soon. If open source can keep up with that is to be seen.

Re: Google “We have no moat, and neither does OpenAI”

#595

Post by hsbauauvhabzb »

narrator wrote: post #396
dahwolf wrote: post #389

The current paradigm is that AI is a destination. A product you go to and interact with. That's not at all how the masses are going to interact with AI in the near future. It's going to be seamlessly integrated into every-day software. In Office/Google docs, at the operating system level (Android), in your graphics editor (Adobe), on major web platforms: search, image search, Youtube, the like. Since Google and other…

I think the problem with AI being everywhere and ubiquitous is that AI is the first technology in a very long time that requires non-trivial compute power. That compute power costs money. This is why you only get a limited number of messages every few hours from GPT4. It simply costs too much to be a ubiquitous technology. For example, the biggest LLama model only runs on an A100 that costs about $15,000 on ebay. The…

Would google even care about integrating LLMs into search? They don’t even prune all the spam entries, presumably because they increase advertising revenue and analytics profit.

Re: Google “We have no moat, and neither does OpenAI”

#596

Post by squiggy22 »

dahwolf wrote: post #389

The current paradigm is that AI is a destination. A product you go to and interact with. That's not at all how the masses are going to interact with AI in the near future. It's going to be seamlessly integrated into every-day software. In Office/Google docs, at the operating system level (Android), in your graphics editor (Adobe), on major web platforms: search, image search, Youtube, the like. Since Google and other…

If Openai can win the developer market with cheap api access and a better product, then distribution becomes through third parties with everyone else becoming the product sending training data back to the model. I'd see that as their current strategy.

Re: Google “We have no moat, and neither does OpenAI”

#597

Post by jacquesm »

safety1st wrote: post #591
dahwolf wrote: post #389

The current paradigm is that AI is a destination. A product you go to and interact with. That's not at all how the masses are going to interact with AI in the near future. It's going to be seamlessly integrated into every-day software. In Office/Google docs, at the operating system level (Android), in your graphics editor (Adobe), on major web platforms: search, image search, Youtube, the like. Since Google and other…

There are no guarantees about who will or won't own the future, just the observation that disruptive technology makes everyone's fate more volatile. Big tech companies like Google have a lot of in-built advantages, but they're notoriously bad at executing on pivots which fundamentally alter or commoditize their core business. If that wasn't true we'd all be using Microsoft phones (or heck, IBM PCs AND phones). In Goo…

> even coming out of the gate pretty strong by dominating the browser market

They were out of the gate about as weak as could be, Windows didn't have a native tcp/ip stack for the longest time (remember Trumpet Winsock?) and they only dominated the browser market through grossly uncompetitive behavior after they had lost the initial 5 rounds of the battle.

Re: Google “We have no moat, and neither does OpenAI”

#598

Post by tlonny »

visarga wrote: post #99
ChaitanyaSai wrote: post #86

This is easily among the rare highest quality articles/comments I've read in the past weeks, perhaps months (on LLMs/AI since that's what I am particularly interested in). And this was for internal consumption before it was made public. Reinforces my recent impression that so much that's being made for public consumption now is shallow and it is hard to find the good stuff. And sadly, increasing so even on HN. As I w…

I've been saying the same things for weeks, right here and in the usual places. Basically - OpenAI will not be able to continue to commercialise chatGPT-3.5, they will have to move to GPT-4 because the open source alternatives will catch up. Their island of exclusivity is shrinking fast. In a few months nobody will want to pay for GPT-4 either when they can have private, cheap equivalents. So GPT-5 it is for OpenAI.…

> I've been saying the same things for weeks, right here and in the usual places. Basically - OpenAI will not be able to continue to commercialise chatGPT-3.5, they will have to move to GPT-4 because the open source alternatives will catch up. Their island of exclusivity is shrinking fast. In a few months nobody will want to pay for GPT-4 either when they can have private, cheap equivalents. So GPT-5 it is for OpenAI.

I wonder if this effect will be compounded by regulatory pressure that seems poised to slow down progress at the bleeding edge of LLMs.

Open source closing the gap at the bottom, and governments restricting further movement at the top...

Re: Google “We have no moat, and neither does OpenAI”

#599

Post by jmiskovic »

> Paradoxically, the one clear winner in all of this is Meta. Because the leaked model was theirs, they have effectively garnered an entire planet's worth of free labor. Since most open source innovation is happening on top of their architecture, there is nothing stopping them from directly incorporating it into their products

I disagree. The model itself is released under GPL3 and no longer "theirs" (Google or OpenAI can use it). And Meta probably has a zoo of such models and I didn't see them use any of the work the community did.

I don't think they "released" LLaMA weights strategically to weaken OpenAI (their overall strategy and market analysis would probably too inert to predict the open source explosion). It probably was a decision by a smaller research team within the company and approved by uninformed executive. Meta could have stepped up and nurtured this small-LLM renaissance, they opted for DMCA hammer instead.

Re: Google “We have no moat, and neither does OpenAI”

#600

Post by seydor »

xkgt wrote: post #528
seydor wrote: post #302

Earlier quoted context omitted.

If i understand correctly it is also shockingly simple, basically just the first figure in the paper: https://miro.medium.com/v2/resize:fit:730/1*D_i25E9dTd_5HMa4... train 2 matrices, add their product to the pretrained weights, and voila! Someone correct me if i m wrong

Correct me if I am wrong, to use LORA fine-tuned model in inference you would still need the original model + trained additional layers, right? If we can perfect methods to fine-tune large models for specific task while reducing the overall model size, then it can fit into more consumer grade hardware for inference and can be broadly used. The objective is to prune unnecessary trivia and memorization artifacts from t…

> to use LORA fine-tuned model in inference you would still need the original model + trained additional layers, right?

You don't need additional layers. After training, the product of the two matrices is added to the original weights matrix, so the model size remains the same as the original during inference.

Post reply on HN