Google Search – Thoughts About & Threats From Microsoft & OpenAI’s ChatGPT

The topic of our conversation is predicated on Google Search and what seems to be growing or intensifying competition with Microsoft and OpenAI. To some extent, Google seems to have announced its Bard offering in response. I think a lot of people probably don’t have a good grasp as to how Google, and maybe Alphabet more broadly, has been investing in and integrating AI and machine learning into search. The company has been doing that for some time. How has it been doing that and could you highlight the magnitude of those efforts?

Google has been investing in things such as RankBrain, neural matching, BERT [Bidirectional Encoder Representations from Transformers] and MUM [Multitask Unified Model]. Some people might not realise that these are technologies the company has invented, pioneered or perfected over the years. Could you speak to some of the specifics that Google has had a major hand in developing and how that’s helped the company differentiate, in terms of the search experience? A lot of people believe it is a leader in search almost by birthright. Its name is synonymous with search, and thus it will be the leader in search forever. While there is some truth to the notion that there is some user inertia that benefits Google, I don’t think people appreciate the magnitude of the investments and innovation in and around search, particularly with an emphasis on AI and machine learning.

I think a lot of people are learning about technologies such as LaMDA [Language Model for Dialogue Applications], something that Google helped pioneer. You seem to have some thoughts on how far ahead of competitors the company may be, whether it’s AI-based search investments or investments in general. Something people may not appreciate is the magnitude and consistency of investments in AI and machine learning it has been undertaking for probably the better part of a decade. I’m not saying that other companies have not been investing, but Google has been investing and rolling out AI technologies, offerings and capabilities. Do you have any thoughts on the amount of money we’re talking about here and how that compares to what else has been spent?

It’s worth mentioning that Alphabet spent USD 40bn on R&D in 2022. Your point seems to be that Google has spent a lot of money on various search and advertising technologies, features, functionality and innovation, and a lot of that is AI and machine learning. This is something that people need to conceptualise, at least in terms of the magnitude of related spending. You mentioned 30,000-50,000 engineers and USD 1m in terms of each one. What do you think Alphabet or Google has spent on AI and machine since it started doing that over the last decade or so? Is it reasonable to think that it has spent USD 100bn on this undertaking?

Importantly, Google has become synonymous not just with search, but, to your point, with trustworthy, reliable and helpful search. When you look at ChatGPT, and what Microsoft and Bing are doing with it, many people ask about the fact that the results are based upon data and information going through, but not past, 2021. The other thing is that it’s not uncommon to go to ChatGPT and try to engage in the platform and it’s not available. My sense is that the reason for these two things is not feasibility but cost. To what extent do you see cost as a major challenge for Microsoft and OpenAI to overcome, as it pertains to rolling out this new offering? Secondly, to what extent do you see them getting past that from a cost perspective and starting to provide the elements that made Google Google in the near or mid-term? It sounds as though it will take years for this to potentially rise to the level that people perceive Google at.

If you were to take the same kinds of requirements and use Google’s infrastructure vs a third-party infrastructure – that is presumably predicated on Azure – what cost differential might there be in the context of what would be necessary to stand up and operate this offering?

It costs a lot of money to operate these offerings. It’s not just the complexity of search and AI and the scale of the user bases, but it would seem that it is going to continue and continue. We seem to be at the dawn of a new era of AI-powered chatbots and search. How does that increase the cost of doing business in this category? It seems you either have to be involved in the category or you’ll be an also-ran. It would seem as though it’s a new cost of business. Is that accurate or overstating it?

You mentioned that the percentage of queries that use conversational AI could reach 5-10% over the near term – and this is not accounting for any real queries at this point – and perhaps as high as 30% or one-third. When might that happen? Is it in 3-5 years?

I think some people would take issue with your comment about Apple not releasing things before they’re perfect – Apple Maps was hardly perfect when that was released. The company has made a lot of improvements since then. Perhaps it learned a thing or two from the experience, and so it’s adhering more to what you were articulating. You also seemed to be suggesting that, yes, Microsoft and OpenAI have the headlines now, but Google and people following Google should perhaps be more concerned about Apple as a competitor, and the implications that Apple is one of the top distributors when it comes to Google Search. That could potentially be a greater risk than Microsoft and OpenAI. Is that a fair characterisation? What time frame would you expect for a proprietary Apple search to be launched? This has been talked about for years.

Meta, when it was called Facebook, launched a Facebook search product that didn’t gain a lot of traction. It’s not lost on me that the company obviously has been investing in it, and I’ve been talking about AI, but perhaps in different ways. I think your point is that Google has been the leader in this category by far for as long as we can remember, and now it seems there is emboldened competition. How might that impact the company? Over the course of this conversation, someone might say, “Avi thinks that Google shouldn’t be as concerned as others think it should be.” On the other hand, you have probably read the same accounts that the launch of ChatGPT in late November 2022 was viewed as a code red by Alphabet and Google leadership. It seems to have lit a fire under them to put together this Bard demo that did not work. I also understand the company is readying a number of AI-related announcements, products and features that could debut at Google I/O in May 2023. It’s not necessarily just third parties, it seems it is focused on and concerned about the developments over the last couple of months. How do you square the notion that this isn’t as big a deal as some people might think with the fact that Google seems to be acting as though it is a pretty big deal?

We’ve been discussing Google Search, AI and what Microsoft and OpenAI are doing. What do you think the next 1-3 years look like? You talked about the possibility of up to 10% of searches having some AI-related aspect, particularly in the answers. What else do you see happening? A lot of people wonder if traditional search will stay separate from this chat-oriented search. Do they come together? Is there an interplay there? Do you see other players getting involved? You highlighted Apple. Do you think it’s reasonable that Apple or Meta could be in this market over the next three years?

Let’s assume that Apple does not change its current relationship with Google and enter as a primary player in the search business. You mentioned that Google has roughly 90% market share. Do you expect that to remain the same, decline or increase in the next three years?

If I said Google might lose 1-2% market share over each of the next three years, do you think that’s an unreasonable assumption? That would get it to mid-80% market share from 90%.

In three years, do you think Apple will be in the search business?