Great overview and I'm loving the supahotfire meme! AI is in an interesting spot right now both with regard to open sourcing it and regulating it. I think open source is generally good for consumers, but I understand why a company wouldn't want to open source AI. I worry that we (as in everyone) don't know enough to properly regulate AI yet but we're still making moves to do so. Or maybe the fact that there is a lot we don't know is the reason to regulate.
The problem to me is that the people regulating are not regulating with good faith. Misinformation and financial interests have muddied the water so much that it's getting hard to sort out the genuine attempts from the smokescreen
The one thing I wanted to push back on is Google's problem just being a PR one. While Google definitely has a broad lead in AI, some other problems Google has:
- Loss of early users because Bard was late in the LLM game: The way I see it, there is not much to distinguish Bard and ChatGPT/Claude etc. as an end user at the moment.
In B2C, ChatGPT has a first to market lead, which resulted in a strong user base that doesn't have a reason to move at the moment. This also helps it drives initiatives like GPTStore more effectively.
In B2B, Bard is still an 'experiment' and for enterprises evaluating AI, many have already started to go with alternatives. Again, earlier to market helps.
A lot has been written about why chatGPT was the first and not Google, so I won't repeat that here.
- TPU game not playing out: Correct me if I'm wrong, my understanding is that one of the reasons TPUs trail GPUs in AI share is that Google kept them all to themself (as a way to potentially make GCP a stronger proposition). The biggest AI compute consumers started with either own compute arrangements (e.g. Meta, Tesla) or tie-ups with large CSPs (e.g. OpenAI). This in turn drove the ecosystem growth around Nvidia which is playing out now. As we start to see more AI compute loads beyond the above, the ecosystem around Nvidia might be hitting GCP in the knuckles.
- No rush to commercialize: Alphabet is a patient parent as far as research goes and there is no aggression/rush to commercialize. Many people speculate this is because of search intoxication. While that has helped in some areas, most notably Waymo, where the patience seems to be resulting in a very strong offering (vav Tesla, Cruise), in other areas it results in ceding markets to more nimble rivals (LLMs).
So while I agree with you that PR is definitely an improvement point for Google, I wouldn't call that the only one. There is some merit to the other coverage too. What do you think?
Hey. I think you might have misunderstood something. This isn't a piece covering the issues at Google, but rather specifically insights into why Google has had to go against OSS in recent times (they have always been one of it's biggest supporters).
I agree completely that Google has more issues than just PR (the main one being a lack of direction and strategy in pushing agendas). However, the reason I point to PR is b/c it specifically is the reason they have now push against OS. Their inability to accurately convey their market positioning and benefits has led to a point where people think every little wrinkle is a challenge to Google's authority (first it was ChatGPT killing Google, now it's Perplexity, even though neither address some of the core issues). Thus they have to lobby against OSS, simply because OSS has the possibility to really hurt public perception.
Working my way through all your incredible insightful posts! I think (if you haven't already done one and I haven't caught up to it yet) you need to give us a perspective on IBM as it seems to rhyme with the Google story (not necessarily from a regulatory standpoint though).
Now that I'm back on my feet (literally) love to catch up (NYC or NJ)!
Well written with excellent graphics and context. The field is going to be full of surprises.
Thank you
Thanks a lot for your work. Please never stop writing!
That is such a kind thing to say. Thank you
Great overview and I'm loving the supahotfire meme! AI is in an interesting spot right now both with regard to open sourcing it and regulating it. I think open source is generally good for consumers, but I understand why a company wouldn't want to open source AI. I worry that we (as in everyone) don't know enough to properly regulate AI yet but we're still making moves to do so. Or maybe the fact that there is a lot we don't know is the reason to regulate.
The problem to me is that the people regulating are not regulating with good faith. Misinformation and financial interests have muddied the water so much that it's getting hard to sort out the genuine attempts from the smokescreen
Very well put. I completely agree.
Thanks Devansh, love this piece :)
The one thing I wanted to push back on is Google's problem just being a PR one. While Google definitely has a broad lead in AI, some other problems Google has:
- Loss of early users because Bard was late in the LLM game: The way I see it, there is not much to distinguish Bard and ChatGPT/Claude etc. as an end user at the moment.
In B2C, ChatGPT has a first to market lead, which resulted in a strong user base that doesn't have a reason to move at the moment. This also helps it drives initiatives like GPTStore more effectively.
In B2B, Bard is still an 'experiment' and for enterprises evaluating AI, many have already started to go with alternatives. Again, earlier to market helps.
A lot has been written about why chatGPT was the first and not Google, so I won't repeat that here.
- TPU game not playing out: Correct me if I'm wrong, my understanding is that one of the reasons TPUs trail GPUs in AI share is that Google kept them all to themself (as a way to potentially make GCP a stronger proposition). The biggest AI compute consumers started with either own compute arrangements (e.g. Meta, Tesla) or tie-ups with large CSPs (e.g. OpenAI). This in turn drove the ecosystem growth around Nvidia which is playing out now. As we start to see more AI compute loads beyond the above, the ecosystem around Nvidia might be hitting GCP in the knuckles.
- No rush to commercialize: Alphabet is a patient parent as far as research goes and there is no aggression/rush to commercialize. Many people speculate this is because of search intoxication. While that has helped in some areas, most notably Waymo, where the patience seems to be resulting in a very strong offering (vav Tesla, Cruise), in other areas it results in ceding markets to more nimble rivals (LLMs).
So while I agree with you that PR is definitely an improvement point for Google, I wouldn't call that the only one. There is some merit to the other coverage too. What do you think?
Hey. I think you might have misunderstood something. This isn't a piece covering the issues at Google, but rather specifically insights into why Google has had to go against OSS in recent times (they have always been one of it's biggest supporters).
I agree completely that Google has more issues than just PR (the main one being a lack of direction and strategy in pushing agendas). However, the reason I point to PR is b/c it specifically is the reason they have now push against OS. Their inability to accurately convey their market positioning and benefits has led to a point where people think every little wrinkle is a challenge to Google's authority (first it was ChatGPT killing Google, now it's Perplexity, even though neither address some of the core issues). Thus they have to lobby against OSS, simply because OSS has the possibility to really hurt public perception.
Working my way through all your incredible insightful posts! I think (if you haven't already done one and I haven't caught up to it yet) you need to give us a perspective on IBM as it seems to rhyme with the Google story (not necessarily from a regulatory standpoint though).
Now that I'm back on my feet (literally) love to catch up (NYC or NJ)!
Let's do NYC at some time. Shoot me a message and we can set something up
What a great topic you have tackled.