The Six Five team discusses Google Gemini Improvements
If you are interested in watching the full episode you can check it out here.
Disclaimer: The Six Five Webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors and we ask that you do not treat us as such.
Transcript:
Daniel Newman: Google’s had its stops and starts, as it’s pertained to LLMs and Gemini, the first version and then they had a bunch of different releases and then they had some people call out that some of the demos weren’t… This time around, at Gemini 2.0 update, I think they did better. I think they were more cautious, they’ve been more careful, and the results so far have been really good. And this is an improvement that’s going to work on your desktop, it’s going to work on your phone. And I think so far we’ve seen some pretty impressive changes that have come with it. First of all, Pat, one of the things that everybody’s up against right now is multimodal. We’re seeing the interest and excitement enthusiasm with OpenAI and Sora. Well, Google is heading down that path with video models, of course with Gemini too, being able to handle more native image processing, audio processing, being able to do process conversion. I mean we are seeing some of the stuff Google is doing with NotebookLM as well. They’re starting to integrate all this stuff. Let’s be candid, I’m starting to see, Pat, some of these notes that Google is going to pull away and maybe OpenAI is in danger. And why is this? It’s because Google has the best video training set on the planet with YouTube. Not to say that OpenAI didn’t train its Sora model with YouTube, but they weren’t supposed to. So that’s going to be a really interesting thing there.
But I see that they’re going to be able to put these things together and be able to really drive some growth and some more positive market sentiment about Google’s play, because Google has been the third or fourth name for a while. I mean it just hasn’t been the top name in LLMs or in AI, despite its AI prowess that I often talk about with the business and the silicon layer.
The other thing is we’re starting to see better integration. And this is one, again, kind of what I was just alluding to, but the integration of Google and all the other apps, all the other data, all of its other ecosystem with what it’s trying to do with LLMs is starting to come together here. So we’re seeing it in their workspace, we’re seeing it, which includes there’s documents and slides and meet, and then of course, Pat, and I’ll kind of turn it over here to you, but the benchmarks, everything just looks better. So there was a number of generation to generation benchmarks and from what I can see between one and then 1.5 and then two is that the performance of the Gemini 2.0 is meaningful and puts it much more parity with some of the other models from OpenAI and Anthropic. And so a lot of improvement in a short period of time, and I expect Google to continue to rise in this space.
Patrick Moorhead: Yeah. Google invented generative AI. They had the seminal research paper on it and somehow, they tripped three times before they brought something out that was relevant and people didn’t make fun of. I mean Bard 1.0, the demo was so bad it was removed, and you cannot find any evidence of that out there. Bard 2.0 was really horrible and it would give you responses and people said there was a lot of political motivations that went into that and they did a Mia Culpa number two, and then they went in and did Google Gemini 1.0 that had (censored) on there.
Daniel Newman: That was awkward.
Patrick Moorhead: I mean give me a break, I mean historically inaccurate stuff and then everybody made fun of them. And my posit was that this really impacted how people looked at not only the developers, all of the developers, could they really trust Google to have a service that would give them relevant advice? And I think they went about six months, nine months without doing something that everybody made fun of. And what really struck me was, again, I didn’t run these market share numbers, it’s from open router AI, but it said that market share amongst developers went from 5% in September to 50% market share last week. And by the way, let’s just say it’s going from five to 25%. That is absolutely monumental. And I think we are starting to see the shift in not only the quality, but the cost and the accuracy of the output that comes out there. Let’s not forget that Sergey, one of the co-founders, essentially came in and started working full time, coming back and motivating troops. And I love, Dan, two years ago we said, “This is going to be a marathon, not a sprint.” I’ve seen some people say that this is where Google makes their move and moves it forward. I’m not there. I think we’re going to see back and forth and competition. We also can’t forget that Microsoft is essentially the leader in all productivity data. It has a very good toehold into enterprise application data. Its Windows servers on-prem hold a tremendous amount of enterprise information. And I don’t see how Google overcomes that at this point. On the consumer side, we’ll see. I’m super impressed even doing Google searches and what these little previews kick out. It’s actually good.
Daniel Newman: It’s super sticky because people are already there.
Patrick Moorhead: I find myself on simple queries doing a little bit more on Google versus starting with Perplexity. So, I’m n equals one and probably, I’m actually signed up for the ChatGPT Pro, right? The 200 bucks a month that I’m trying out. I don’t consider myself a normal user, but at a minimum, the winds are pushing in the Google direction here.
Daniel Newman: Yeah. No, you hit it on the head. I think a lot of progress has been made. I mean, I don’t know about you. I’m getting a ton of media calls about the possible breakup and the spin of Chrome and what’s going to happen with Google and generative AI is going to completely upend search. So the question is, can Google maintain its leadership? And I mean I think the progress they’ve made, being able to put generative inside of search, it’s powerful. I mean we certainly didn’t see Bing’s ChatGPT integration move market share, it just didn’t happen. So they have a good moat, but we have seen how fast ChatGPT has grown.
