> When I searched my name, there it was: a picture of my smiling face next to the text “Tom Faber was a physicist and publisher, and he was a university lecturer at Cambridge for 35 years”.
Good grief. Pal, you've got a name with 8 identifying alphanumeric characters. Of course it is going to collide with other identities on the openly-indexed internet.
If it were a simple name collision in the way the original commenter seemed to mean, you’d search for your name, and see the photo and biography of a different person.
The article writer searched for his name, and found his photo attached to somebody else’s biography. That’s an incorrect search result.
No, he found another person's biography near his profile photo for his X blurb. They were in distinctly different content bubbles, even if it visually looked similar.
To me, a name collision is when two or more entities share the same name or ID. What's claimed here is that Google was showing the photo of one person on the bio of another person. That's an identity consolidation error, where two unrelated people are treated as one. The name collision presumably caused the identity consolidation error, but you can have name collisions without identity consolidation errors.
So Google is presenting summary information in a way that very much implies comprehension, but without any comprehension of context — i.e. a name collision is enough to throw them off and result in unrelated pictures and bio being smashed together?
Seems pretty irresponsible of Google to do that. And also, just... kind of lame. Google is slowly but surely changing their image from "tech wizard" to "kind of a joke", and they have nobody but themselves to blame.
It's AI-generated, like anything AI-generated you should not assume it's totally accurate. I would support some kind of disclaimer if it's not already there.
When I search for "Tom Faber", I expect to get results relating to a bunch of Tom Fabers.
When I see a knowledge panel result, I expect everything within it to relate to a single entity, regardless of name collisions. If it doesn't, that's incorrect. If it's too hard to disambiguate, it's better to not have the knowledge panel at all.
That said, I did this search just now. https://imgur.com/a/zEjrfte The top organic result is the author's X bio and has his profile pic in the search results page. To the right of it (across a gutter) is the knowledge panel about the physicist. I could see how one might get confused, but the X profile pic is just near the knowledge panel, not actually in it. So if this is what the author saw, I don't agree with the article. It'd be nice if he included an actual screenshot of what he's describing, instead of the cute "Why is Google..." thing.
> When I see a knowledge panel result, I expect everything within it to relate to a single entity, regardless of name collisions. If it doesn't, that's incorrect.
Sure, that's the goal. But it's not like this is an unreasonable thing to get wrong on Google's part. It's hard, this is not like they're failing at easy tasks here.
"Sure, that's the goal. But it's not like this is an unreasonable thing to get wrong on Google's part. It's hard, this is not like they're failing at easy tasks here."
Should unreasonable trump acceptable? If you fail at something then should you do it?
An uncharitable interpretation seems to allow it is acceptable to imply someone is dead because an algorithmic process is hard ... lol ... l8ers.
So we shouldn't use AI at all because it's often inaccurate? I think it's fine as long as it's made clear that it's not totally reliable, both in widespread public knowledge and in UI disclaimers.
I reckon it's reasonable to say we shouldn't use AI (LLMs) in places where we're looking for accurate information, yeah. Expectations for accuracy are higher in a seemingly curated summary.
Right, which is why reasonable levels of humility and wisdom would keep an organization from expanding their mission from indexing documents to providing automated answers to questions.
Yes, it's likely an unmitigable problem with trying to aggregate information in this way. That's why the value proposition of a search engine is that it presents information from primary sources, rather than trying to summarize it. Finding resources on other people with the same name is unsurprising, having the AI conflate you as it presents an aggregate is misleading, and enough so to be a real problem
> Google provides a feedback form to resolve this type of bug. I filled it in several times, but it made no difference.
At this point they're certainly allowed to be surprised, annoyed (pick whatever word you want to describe it) that it's still happening. No need to be snarky, pal.
Was just using OP's own word choice ("pal") to highlight the unnecessariness of it. Given the current color of your comment, it seems like others have correctly picked up on what I was doing there. :)
I read the article and what I understood that a "knowledge panel/info box" popped up with this living man's picutre and a note saying he was deceased. It wasn't just an issue of several results coming up for a person with a common name.
I once switched to Google, one day in '98, when they were basically ad-free, with good results.
I once switched away from Google, one day in '23, when the first seven results for many queries were 'sponsored', and shitty.
These days, I let LLMs answer most of my questions. They are sufficiently easy to use. For regular surfing, I rely on Kagi. Both services are pay for play, but my lifetime is not unlimited, and can be spent better than shifting through advertisements.
You've got to be careful about which LLM. I asked Microsoft Copilot for advice about staying in Paris, and then as a followup in the same thread, asked about navigation apps.
It started recommending apps that work well... in New York City.
If Copilot can't retain context, in the same thread, after a single message, then it's useless.
ChatGPT, which apparently Copilot is based on, didn't have this problem.
And from the linked article I have just discovered "Perplexity", which looks a bit like ChatGPT only plugged into a search engine. ChatGPT is also plugged into a search engine, so maybe it's plugged in... more? Not sure.
If anyone working at Perplexity is reading this, please add an "about" page to tell me what the site actually does. Nowhere I could find would explain. Everything assumes you already know, and I eventually had to go to Wikipedia to find out.
If anyone makes an LLM which only knows what's in Wikipedia, and doesn't hallucinate anything else, they would clean up.
Because "searching for an encyclopedia article" and "asking questions about a branch of knowledge" are fundamentally different approaches. In theory, an LLM that was a talking Wikipedia would not be limited to giving you knowledge from just one page, it could extract multiple pieces of information from multiple articles to answer the precise question you had, and could accept follow-up questions on the same or similar topics.
It's a good idea, but it might turn out to hallucinate even more. For instance, there is one article on Abraham Lincoln and one on Abraham Lincoln: Vampire Hunter, so the statistics (50/50) in regard of mixing fantasy with reality are perhaps discouraging.
If you know about the subject LLMs makes you turbo. Even when you don't know about the subject a rational thinking background gives you more legs.
LLMs are incredible tools, it is expected that they are not the only ones in your toolbox.
When I take a random photo and asks ChatGPT to describe it I always wonder how I saved a whole lifetime (or many ones) dedicated to computer vision research and can build upon all of this.
> ChatGPT is also plugged into a search engine, so maybe it's plugged in... more? Not sure.
I haven't used ChatGPT search, but I have used Perplexity. The way that it works, it will potentially do like a dozen searches to try to understand your question (if it needs to), search individual topics in the question, search joint topics in the question, and then take 20 results or so, including web pages and videos, and synthesize an answer to your question.
> If anyone makes an LLM which only knows what's in Wikipedia, and doesn't hallucinate anything else, they would clean up.
Kagi has a "lens" feature that lets you scope down web results to certain sources. It's supposed to be able to be used in their AI assistant, but it's currently ignored / broken there. They are working on a large overhaul (I think it's supposed to be out by the end of this month?), hopefully that works after that. You could achieve what you described using that then.
LLMs are also capable of mixing up information about different people:
> Prompt: Who is Tom Faber
> ChatGPT: Tom Faber is a versatile individual with multiple professional identities. He is known as a culture journalist and gaming reviewer, contributing to prestigious publications like the Financial Times, The Guardian, and The New York Times. His work often covers topics related to arts and gaming. Additionally, Tom Faber serves as the Clinical Director of the Zuni Indian Health Service Hospital and is a founding member of the Zuni Youth Enrichment Project (ZYEP), where he has been instrumental in developing programs for the Zuni community since 2006 (Zuni Enrichment) .
It's probably obvious here that it's talking about two different people as the fields are unconnected, but if they were closer, you'd assume it's the same person.
They can also invent things:
> Prompt (new chat): tell me about tom faber
> Tom Faber is a prominent figure in British literature, known for his work as an author and editor. His most notable works include his novel "The Scent of Death," which delves into themes of mystery and psychological drama. Faber's writing often features intricate plots and rich character development, reflecting his deep understanding of human nature and storytelling.
The only book titled "The Scent of Death" I can find was written by Simon Beckett.
But when Google was young and wonderful, the web was a relative paradise, and rich with truthful (or at least well-intended) information.
Now, the web is mostly an SEO-optimized, Terminal-Stage Capitalist hellscape. Even the very best talents and intentions would find it extremely difficult to create a "good" search engine these days.
I mostly use non-Google search now. Even when it ain't better, I can imagine that I'm boosting the "people looking for less-sucky-than-Google search" stats.
I don't really think that's true. I think most of the web you find from Google is mostly SEO-optimized. There are a lot of web pages. And a lot of them, Google does not promote.
I have had a really good experience with web search since I switched to Kagi last year. Like, I've genuinely been able to find things that I have been unable to find with other search engines.
Yes, and it’s been broken for a long time. The crossroads of siting on one’s laurels and conflict of interest have resulted in Google’s results not being particularly good.
The only reason they remain on top (IMO) is that the dynamics of search, how information is published, and how people discover new information are rapidly changing.
I somewhat disagree. By way of example, when I search for "unreal engine UObject" I would expect to get the documentation page for Unreal Engine's UObject class. (https://dev.epicgames.com/documentation/en-us/unreal-engine/...) Google search was at one point excellent for this sort of thing! But, today, when I search for that, the UObject documentation page is nowhere to be found.
It is still OK for Python stuff. But it used to be good for everything!
That's a manifestation of fear at management level. They fear punishment from Wall Street for not implementing an AI strategy so they force their teams to "do something", with predictable results, i.e. garbage produces by LLMs indiscriminately recombining data.
There is plenty of great information that google is just unable or unwilling to return these days which it returned just fine in yesteryears.
I'm not talking about political bullshit or culture whatever, though Alphabet has serious problems there too, but just boring stuff like technical documentation that doesn't even have a lot of obvious spam interference.
This implies that someone else could just do what Google used to do, create a better search engine, and eat their lunch. They might not make as much money per-user, but for all search engines other than Google, their biggest problem is not revenue per user, it's how to get more users. "Have a better search engine" is a pretty good advantage there.
I use kagi exclusively now except for shopping and images. Anytime I use Google, now I'm disgusted with how bad the results are. The bar seems really low to supplant Google search.
The secondary problem here - one that is both larger and not something Kagi can solve - is that so much content on the internet is just low quality garbage. Even supposedly crowdsourced info from Reddit is astroturfed to hell by social media marketing firms.
So what good is it finding something that’s highly relevant to my search query if it’s just garbage anyways. There are just so few sources of meaningful information, and essentially no way to reliably differentiate them from drivel.
I depended a lot on Google search. Since Google start moving to the dark side, I was wondering when could I remove the dependency. It turned out the dependency removed itself.
Yes, definitely. This is the reason the summary box is telling you to eat glue - that’s what the top search results say, and the LLM dutifully summarizes.
Google has fucked up search so badly it is fucking up their AI ambitions. Very karmic.
It’s probably basically google-bombing for the modern era underneath. But the point is that search is now so broken that it can be pretty trivially google-bombed in 2024, at the exact moment the world needs it for RAG.
> Nayak apologised, saying the panels were created automatically using algorithms and sometimes they messed up: “These are the kinds of things we’re constantly improving.”
A process or invention may be "constantly improving" by small and meaningless amounts without ever reaching a point that offsets the damage it does over its lifetime.
"Our toaster now uses 5% less asbestos, and we're passing on the savings to you!"
I renovated and reopened an old theater in San Francisco years ago. Google maps still lists my theater as “Chinatown's 1925 cinematic home to kung fu & opera films is closed but reopens for film festivals.” Which has not been accurate for a decade. I’ve claimed the location and updated everything I can of the Google maps description but this seems editorialized and there’s no place to edit it or mark it as invalid. I’ve tried reporting the listing multiple times but there’s no way to get a response from a human or get any changes made.
I know it’s not that related to the article but Google is broken and so many little ways, it’s not surprising it adds up to a slowly crumbling quality overall.
That sounds so annoying. Could it possibly be something they source upstream from somewhere else on the internet?
I.e if you search verbatim "Chinatown's 1925 cinematic home to kung fu & opera films is closed but reopens for film festivals.", there are two websites that list your business with that exact description. Either they got it from Google, or Google got it from them. It might be worth claiming it on those sites and changing as well in the off chance it's the latter.
I have a relatively uncommon name. Way back in the beginning Google thought I was an 18th century English criminal. Then for a while in the 2010s I gave a shit about SEO and got my own website ranked first for my name. Then I quit caring and last time I checked, Google thinks I'm an 18th century criminal again.
That's the same fallacy as the author of the article: No, Google doesnt “think you are an 18th century English criminal”, any more than it thinks the author “was a dead physicist”. It knows that he, the 18th century English criminal, was an 18th century English criminal. Just like it knows that the other guy, the dead physicist, was a dead physicist. It isnt talking about you (or the OP author); it's not as if you two are the only ones ever to bear your names.
73 comments
[ 6.1 ms ] story [ 165 ms ] threadGood grief. Pal, you've got a name with 8 identifying alphanumeric characters. Of course it is going to collide with other identities on the openly-indexed internet.
> confused pictures of my face with the biography of another man who shared my name
The article writer searched for his name, and found his photo attached to somebody else’s biography. That’s an incorrect search result.
Seems pretty irresponsible of Google to do that. And also, just... kind of lame. Google is slowly but surely changing their image from "tech wizard" to "kind of a joke", and they have nobody but themselves to blame.
When I see a knowledge panel result, I expect everything within it to relate to a single entity, regardless of name collisions. If it doesn't, that's incorrect. If it's too hard to disambiguate, it's better to not have the knowledge panel at all.
That said, I did this search just now. https://imgur.com/a/zEjrfte The top organic result is the author's X bio and has his profile pic in the search results page. To the right of it (across a gutter) is the knowledge panel about the physicist. I could see how one might get confused, but the X profile pic is just near the knowledge panel, not actually in it. So if this is what the author saw, I don't agree with the article. It'd be nice if he included an actual screenshot of what he's describing, instead of the cute "Why is Google..." thing.
Sure, that's the goal. But it's not like this is an unreasonable thing to get wrong on Google's part. It's hard, this is not like they're failing at easy tasks here.
Should unreasonable trump acceptable? If you fail at something then should you do it?
An uncharitable interpretation seems to allow it is acceptable to imply someone is dead because an algorithmic process is hard ... lol ... l8ers.
At which point is disinformation acceptable?
Any form of auto-generated summary online is from AI
Preferably not, yeah.
At this point they're certainly allowed to be surprised, annoyed (pick whatever word you want to describe it) that it's still happening. No need to be snarky, pal.
Interesting time to choose to be snarky.
I once switched away from Google, one day in '23, when the first seven results for many queries were 'sponsored', and shitty.
These days, I let LLMs answer most of my questions. They are sufficiently easy to use. For regular surfing, I rely on Kagi. Both services are pay for play, but my lifetime is not unlimited, and can be spent better than shifting through advertisements.
It started recommending apps that work well... in New York City.
If Copilot can't retain context, in the same thread, after a single message, then it's useless.
ChatGPT, which apparently Copilot is based on, didn't have this problem.
And from the linked article I have just discovered "Perplexity", which looks a bit like ChatGPT only plugged into a search engine. ChatGPT is also plugged into a search engine, so maybe it's plugged in... more? Not sure.
If anyone working at Perplexity is reading this, please add an "about" page to tell me what the site actually does. Nowhere I could find would explain. Everything assumes you already know, and I eventually had to go to Wikipedia to find out.
If anyone makes an LLM which only knows what's in Wikipedia, and doesn't hallucinate anything else, they would clean up.
LLMs are incredible tools, it is expected that they are not the only ones in your toolbox.
When I take a random photo and asks ChatGPT to describe it I always wonder how I saved a whole lifetime (or many ones) dedicated to computer vision research and can build upon all of this.
I haven't used ChatGPT search, but I have used Perplexity. The way that it works, it will potentially do like a dozen searches to try to understand your question (if it needs to), search individual topics in the question, search joint topics in the question, and then take 20 results or so, including web pages and videos, and synthesize an answer to your question.
> If anyone makes an LLM which only knows what's in Wikipedia, and doesn't hallucinate anything else, they would clean up.
Kagi has a "lens" feature that lets you scope down web results to certain sources. It's supposed to be able to be used in their AI assistant, but it's currently ignored / broken there. They are working on a large overhaul (I think it's supposed to be out by the end of this month?), hopefully that works after that. You could achieve what you described using that then.
> Prompt: Who is Tom Faber
> ChatGPT: Tom Faber is a versatile individual with multiple professional identities. He is known as a culture journalist and gaming reviewer, contributing to prestigious publications like the Financial Times, The Guardian, and The New York Times. His work often covers topics related to arts and gaming. Additionally, Tom Faber serves as the Clinical Director of the Zuni Indian Health Service Hospital and is a founding member of the Zuni Youth Enrichment Project (ZYEP), where he has been instrumental in developing programs for the Zuni community since 2006 (Zuni Enrichment) .
It's probably obvious here that it's talking about two different people as the fields are unconnected, but if they were closer, you'd assume it's the same person.
They can also invent things:
> Prompt (new chat): tell me about tom faber
> Tom Faber is a prominent figure in British literature, known for his work as an author and editor. His most notable works include his novel "The Scent of Death," which delves into themes of mystery and psychological drama. Faber's writing often features intricate plots and rich character development, reflecting his deep understanding of human nature and storytelling.
The only book titled "The Scent of Death" I can find was written by Simon Beckett.
That is a great route to being duped by LLM hallucinations, but not so much for actual knowledge.
But when Google was young and wonderful, the web was a relative paradise, and rich with truthful (or at least well-intended) information.
Now, the web is mostly an SEO-optimized, Terminal-Stage Capitalist hellscape. Even the very best talents and intentions would find it extremely difficult to create a "good" search engine these days.
I mostly use non-Google search now. Even when it ain't better, I can imagine that I'm boosting the "people looking for less-sucky-than-Google search" stats.
I don't really think that's true. I think most of the web you find from Google is mostly SEO-optimized. There are a lot of web pages. And a lot of them, Google does not promote.
I have had a really good experience with web search since I switched to Kagi last year. Like, I've genuinely been able to find things that I have been unable to find with other search engines.
The only reason they remain on top (IMO) is that the dynamics of search, how information is published, and how people discover new information are rapidly changing.
It is still OK for Python stuff. But it used to be good for everything!
That wholly sounds like a Google problem to me.
I'm not talking about political bullshit or culture whatever, though Alphabet has serious problems there too, but just boring stuff like technical documentation that doesn't even have a lot of obvious spam interference.
https://www.wheresyoured.at/the-men-who-killed-google/
So what good is it finding something that’s highly relevant to my search query if it’s just garbage anyways. There are just so few sources of meaningful information, and essentially no way to reliably differentiate them from drivel.
But you're still right that there's so much garbage out there. Google had it largely under control though.
Google has fucked up search so badly it is fucking up their AI ambitions. Very karmic.
It’s probably basically google-bombing for the modern era underneath. But the point is that search is now so broken that it can be pretty trivially google-bombed in 2024, at the exact moment the world needs it for RAG.
A process or invention may be "constantly improving" by small and meaningless amounts without ever reaching a point that offsets the damage it does over its lifetime.
"Our toaster now uses 5% less asbestos, and we're passing on the savings to you!"
I know it’s not that related to the article but Google is broken and so many little ways, it’s not surprising it adds up to a slowly crumbling quality overall.
I.e if you search verbatim "Chinatown's 1925 cinematic home to kung fu & opera films is closed but reopens for film festivals.", there are two websites that list your business with that exact description. Either they got it from Google, or Google got it from them. It might be worth claiming it on those sites and changing as well in the off chance it's the latter.