Paul Murphy Challenges Google on AI 'Hallucinations' in Search

Thanks a lot for coming before us. Just to start, I mean, I presume you don't accept this, but in my perception that Google has gotten worse at producing the basic thing of what Google does or traditionally has done, which is we want to search for something, we go on Google, we look for it. Because what happens now if we search most things on Google is that we get some AI-generated stuff. I mean, AI-generated stuff we heard last week. The purpose is not to tell the truth. That's not what it aims to do. LLMs are simply predicting the most likely next word. So the whole thing is a hallucination. Sometimes those hallucinations are accurate, often they're accurate, but often that they're not accurate. Do you agree that effectively Google is participating in what's being termed the initiatification of the internet? Let me take your question on AI overviews, which is how we bring AI into search. It works differently from traditional LLMs. So it is trained and grounded on the core search ranking feature. So that is our 25 years of experience in elevating high-quality content to users. And so when users come to ask these increasingly complex questions, we provide them in a summary. We provide the users with links to those sites. And what we are finding, in fact, is that users really like this tool and find that it is valuable. So that when they click over to the websites, it's a high-quality click. And we know that because they are spending more time on that site and not clicking back right away. So we are seeing improvements and bringing our tools to where users are. I mean, what about the recommendation that you use non-toxic blue to stick cheese on a pizza? So those can be, you know, we've seen those headlines as well, certainly. In reality, those kinds of queries are very rare and those kinds of results are very rare. But certainly when we see them, we work hard to make changes and address it at not just the level of that query, but at a more systemic level. Or the invention of a one kilo tomato for 1,000 kilometres, a new unit of measurement. I think that's another good example of a rare query, but one that when it comes up... Or the suggestion that recommending eating at least one small rock per day for minerals. Similarly, and what you're pointing out is that there are some sites for comedy or forums for farce where these can pull from, and that's not the result we would want to see. Again, very rare, and when we see it, we work to correct it. There's a paper out recently by Google researchers, apparently, which has concluded that generative AI is ruining vast swathes of the internet with fake content. And the distinction between what is real and what is fake is increasingly blurred. Do you accept that Google is engaged in this process? I mean, you recently launched a video-generating AI where you can generate from a prompt a seven or eight seconds video. Do you accept that this is, like, undermining the distinction between truth and reality and is assisting in the proliferation of fake news, spread of hate, and so on? So, you know, we think these are incredibly valuable tools, and I'm not familiar with that paper. I've read a lot of the Google research papers, and that sounds like an interesting one. But what I would say is Google has, again, many, many years of experience in addressing low-quality spam content on search, on YouTube, and across platforms. And what this new generation of AI technology can bring is immense creativity, incredible new pieces of content, and it can also bring the other kind. And for that, we will be relying on our many years of experience here. Thanks. Thank you. Kate?