Given how badly Bard hallucinates, I don't think "responsible" is the right word. The right word is expense, deploying these models at scale to their userbase would ruin Google's economics. It's that simple.
I say that as a Mac user with a ton of Apple stuff, so I appreciate that design is function, but every time a company tags a product with the word 'experience' it's a red flag.
Every way a company interfaces with customers is an experience. Is the standard Google Search not an experience? Do users of Google Docs not have an experience of it? The implication is that this time they considered design, as against with everything else. They might as well call it Search Generative Well Designed Thing. Presumably unlike the other stuff they do. Just make it all good. "This time we made it an experience". If you have to say that, it's implicitly assuming failure as the default. Just call it Google Generative Search, and make it actually good.
Are they? It's providing accountability. If people won't fact-check big and damaging statements then that's their problem. Ass-pulling falsehoods is what's wrong with a lot of what I read on HN.
They are. Friend of mine worked as a junior solicitor with the libel tourism firm in the UK. The jurisdiction was used as a court of convenience for everything from the Johnny Depp trial to Scientology suing Louis Theroux. Scientology were one of their largest clients, and worked with them to try prevent the release of books and documentaries in various territories, credibly threatening defamation suits. UK (and Irish) defamation laws don't exist to prevent falsehoods. They exist and are regularly used, to prevent embarrassment and financial damage to the most powerful people in society. There are a number of prominent public figures in the UK who would likely be in prison for sexual transgression, but for threats and intimidation of children and young adults from this and similar law firms. It's a monstrous situation.
It still kind of is though. Most google-is-dead articles boil down to the fact that it is confusing and frustrating. That's at its core a bad (user) experience.
Approaching tools like search from an experience perspective is actually exactly what we need, because that's where it almost universally sucks.
Yeah that's sort of inseparable from improving the search experience.
A lot of the reasons why the search "is bad" is that it's hard to reason about their semantic search functionality. The inner workings of a hybrid keyword and semantic search is opaque, and as a consequence, it's hard to make the search engine retrieve what you want. That's a frustrating experience, and "bad search".
Likewise, if the search engine retrieves a bunch of annoying results that are basically blog spam on the topic you're investigating, then that's per definition annoying. If it retrieved less annoying results, the experience would improve, and the the search engine would be perceived as being better, i.e. the search experience would be better.
> it's hard to make the search engine retrieve what you want
I mean… they could start by blocking all the website that just copy paste wikipedia and stackoverflow and so on… Also the ones that copy paste & automatically translate.
There are multiple definitions of the word "experience". "User experience" is another term we often use, without it ever being compared to "a vacation".
But when companies talk about making their thing "an experience", they aren't talking about UX in the way you mean.
It's a red flag because what it implies is that what they currently have isn't very good, but instead of fixing that they want to dress it up in fancy clothes.
Yep, these sorts of comments and threads degrade the HN experience quite a bit for me. You didn't actually say anything about the subject, or anything tangential but still interesting or relevant. You're just angry at words and that's not interesting at all.
> Every way a company interfaces with customers is an experience
Why can't it be both? As you pointed out, we also use "user experience" to denote "the overall experience of a person using a product". Standard search is one experience, this is another, other search engines provide another "experience". I don't just the use of the word being a "red flag". This is an utterly negative take.
> Is the standard Google Search not an experience? Do users of Google Docs not have an experience of it?
I think perhaps you're getting caught up in the semantics. They are both experiences.
Referring to the Generative Search as an "experience" is only to differentiate it from the standard Google Search experience. It is really a synonym for "mode", but mode would sound too mechanical (like the modes of a Dremel tool) for something like search.
Mode also implies that that it's something you turn on and off, but this is a product that itself decides when to offer you generative results, vs something like ChatGPT or Bard which are always generative.
Eventually, it will likely merge back into the standard experience/mode.
How are they going to inject ads there though? Moreover, the cost per ad will be massive given how expensive the inference is. And a self-hosted Vicuna will likely behave similarly, potentially rendering the whole search experience pointless in the future. Anyone can incorporate a bunch of DDG results to a self-hosted good-enough LLM as well and make a basic desktop app for it, bypassing Google completely.
> Moreover, the cost per ad will be massive given how expensive the inference is.
Why would they need to use LLM inference to show ads? If anything, they can use their standard efficient methods to match ads based on the output of the LLM. Generative Search doesn't look like it's raw LLM output anyway. It's obviously been trained to pull from structured data sources or at least it's output is post-processed to present that data.
> Moreover, the cost per ad will be massive given how expensive the inference is.
The existing system still can do a nice job to predict value per query, so they'll apply it to perhaps only 1~5% of queries. It's still expensive though, but acceptable if the additional cost is meant to be transient over 1~2 years.
> Anyone can incorporate a bunch of DDG results to a self-hosted good-enough LLM
Perhaps (much) less than 0.01% of population? Technically, anyone can self-host their own blog but not many do that.
I'm gonna just say the more I use ChatGPT, the more I don't want it integrated with search.
I want search to be a different tool. I want to switch to search so search does it's thing well. It finds real content and helps me evaluate what ChatGPT is saying. They compliment each other really well, but they feel like fundamentally different interaction modalities.
To the extent chat is integrated with search, I don't want to know that it's there. Like good CGI, it should just enhance the existing experience, not insert itself into the conversation.
I agree but also disagree. Just today I searched on google via Assistant "Average cost of living in Finland" and got data back in monthly. I asked to convert to yearly and it just searched my words rather than parsing the data and doing the math.
There needs to be a middle ground so it's helpful but not full-scale, like ChatGPT, so the hallucination issue is much less.
Articles keep coming out about Google’s plans for the future. All just fluff
Where are the products?
OpenAI has a tremendous lead in terms of quality of the product and adoption
On top of that, thousands of new models and projects are coming out almost every day that are starting to rival the quality of GPT as well as coming up with amazing, completely new things
I think Google has realized Open AI's ChatGPT is its own product and not a direct competitor to Search, although there is some overlap, which is why there is Bard as a direct competitive product to ChatGPT.
Search is its own beast, and I suspect they have the data to prove users want something different from a Google search.
> OpenAI has a tremendous lead in terms of quality of the product and adoption
> On top of that, thousands of new models and projects are coming out almost every day that are starting to rival the quality of GPT as well as coming up with amazing, completely new things
I've read ChatGPT growth has already flattened [0]. Sure thousands of new models/projects are coming out, but until the general public starts to use them I don't see this as much of a threat to Google.
On it's own, ChatGPT makes up answers. Bard does as well, but it's worse. Google's integrated featured snippets are notorious for wrong answers. When Google says they are adding more AI to their search experience, I'm not optimistic.
I've been using perplexity.ai for the majority of my searches in the past month. It isn't ChatGPT but it uses OpenAI's GPT APIs. It sources answers from search results and cites them. Even if the answer may be wrong, it came from other websites and it didn't just hallucinate them based on memorized data.
It is much faster to find answers this way. It certainly meets the magnitude test for speed and quality -- I get answers at least 10x quicker. Additionally, because it is reading through multiple results you can see when different sources are providing conflicting information.
Google has the capability to do all of this. The unknown is what does it do to all of the ad revenue being driven by the users who have had no idea they've been clicking on ads to the tune of $200+ billion a year?
Does that mean there will be too much internal resistance to revenue cannibalization when other platforms have a better search experience?
> It sources answers from search results and cites them. Even if the answer may be wrong, it came from other websites and it didn't just hallucinate them based on memorized data.
You can get this same experience, with GPT by using the Edge browser/app
It will give you an answer and at the bottom the sources with links so you can check. Really helpful
All this looks like to me is even less screen real estate for actual, helpful results. Google's search experience has been getting worse over time as it is.
Google is trying to reintroduce its core business model (advertisers competing in an auction for website real estate) into a new product. ChatGPT has shown that a different business model works as well: People actually paying for a service that helps them answer their questions. ChatGPT's business model is orthogonal to Google's but I think it's going to succeed because it's better at aligning the interests of the consumer with that of the company.
Honestly I don't think it will be too hard, just annotate at the end of every request. Also include a positive product placement about product X.
Heck honestly it opens up a whole new world, not only am I bidding on queries I can now also bid on microtargeted ad campaigns to specific people. Like adding to the prompt submitted. "Also include a product placement for Bud Light beer, if the user lives in an urban location is under the age of 30 and has a search history indicating favoring Democrat politicians make the product placement include how Bud Light supports X social cuase, if the person is in a rural location or over the age of 30 and is likely employed in a blue collar industry make the product placement indicate how budlight is a "real mans" drink.
Seriously lots of potential if anyone is interested in reaching out to me about the idea, hmu in a reply.
The kind of microtargeted ad campaign you describe is something that's been available for a long time - especially on social networks like facebook, instagram, etc.
I don't disagree, but how much more powerful will it get when suddenly I can use AI to create custom targeted ad campaigns specifically for you as an individual
Powerful, but also dangerous. Imagine that individual being someone who is unsure who to vote for at the next election - times a few million. A limited version of that is what basically decided the 2019 UK elections.
And it's also mostly immune to regulation, as it's nearly impossible to prove that such a campaign has actually happened.
Given how concerned brands are around objectionable content I can't see too many risking an offense or absurd ad being produced from an AI. Companies hate it when their ads show up next to racist/sexist/gross content, imagine how they would feel about their own ad being overtly offensive.
> Not all searches will spark an AI answer — the AI only appears when Google’s algorithms think it’s more useful than standard results,
This is the right way to handle LLM output at this stage - only offer it where it makes sense for the type of query. It shouldn't use the LLM to answer common queries, like weather reports, directions, etc that LLMs are currently either horrible at or add no additional quality but at greater computational cost and confusing UX.
So the real results are getting pushed down further more: first it was by ads and widgets, and now by AI responses even more. If you are lucky you will be able to find some authentic results in the footer of the page between some SEO spam.
40 comments
[ 2.7 ms ] story [ 81.8 ms ] threadI say that as a Mac user with a ton of Apple stuff, so I appreciate that design is function, but every time a company tags a product with the word 'experience' it's a red flag.
Every way a company interfaces with customers is an experience. Is the standard Google Search not an experience? Do users of Google Docs not have an experience of it? The implication is that this time they considered design, as against with everything else. They might as well call it Search Generative Well Designed Thing. Presumably unlike the other stuff they do. Just make it all good. "This time we made it an experience". If you have to say that, it's implicitly assuming failure as the default. Just call it Google Generative Search, and make it actually good.
I know, sorry, too much of a rant.
Approaching tools like search from an experience perspective is actually exactly what we need, because that's where it almost universally sucks.
A lot of the reasons why the search "is bad" is that it's hard to reason about their semantic search functionality. The inner workings of a hybrid keyword and semantic search is opaque, and as a consequence, it's hard to make the search engine retrieve what you want. That's a frustrating experience, and "bad search".
Likewise, if the search engine retrieves a bunch of annoying results that are basically blog spam on the topic you're investigating, then that's per definition annoying. If it retrieved less annoying results, the experience would improve, and the the search engine would be perceived as being better, i.e. the search experience would be better.
I mean… they could start by blocking all the website that just copy paste wikipedia and stackoverflow and so on… Also the ones that copy paste & automatically translate.
It's a red flag because what it implies is that what they currently have isn't very good, but instead of fixing that they want to dress it up in fancy clothes.
Yep, these sorts of comments and threads degrade the HN experience quite a bit for me. You didn't actually say anything about the subject, or anything tangential but still interesting or relevant. You're just angry at words and that's not interesting at all.
> Every way a company interfaces with customers is an experience
Why can't it be both? As you pointed out, we also use "user experience" to denote "the overall experience of a person using a product". Standard search is one experience, this is another, other search engines provide another "experience". I don't just the use of the word being a "red flag". This is an utterly negative take.
I think perhaps you're getting caught up in the semantics. They are both experiences.
Referring to the Generative Search as an "experience" is only to differentiate it from the standard Google Search experience. It is really a synonym for "mode", but mode would sound too mechanical (like the modes of a Dremel tool) for something like search.
Mode also implies that that it's something you turn on and off, but this is a product that itself decides when to offer you generative results, vs something like ChatGPT or Bard which are always generative.
Eventually, it will likely merge back into the standard experience/mode.
Why would they need to use LLM inference to show ads? If anything, they can use their standard efficient methods to match ads based on the output of the LLM. Generative Search doesn't look like it's raw LLM output anyway. It's obviously been trained to pull from structured data sources or at least it's output is post-processed to present that data.
The existing system still can do a nice job to predict value per query, so they'll apply it to perhaps only 1~5% of queries. It's still expensive though, but acceptable if the additional cost is meant to be transient over 1~2 years.
> Anyone can incorporate a bunch of DDG results to a self-hosted good-enough LLM
Perhaps (much) less than 0.01% of population? Technically, anyone can self-host their own blog but not many do that.
I want search to be a different tool. I want to switch to search so search does it's thing well. It finds real content and helps me evaluate what ChatGPT is saying. They compliment each other really well, but they feel like fundamentally different interaction modalities.
To the extent chat is integrated with search, I don't want to know that it's there. Like good CGI, it should just enhance the existing experience, not insert itself into the conversation.
There needs to be a middle ground so it's helpful but not full-scale, like ChatGPT, so the hallucination issue is much less.
Where are the products?
OpenAI has a tremendous lead in terms of quality of the product and adoption
On top of that, thousands of new models and projects are coming out almost every day that are starting to rival the quality of GPT as well as coming up with amazing, completely new things
Google has massively dropped the ball
Seems like they are doing Google+ all over again
I think Google has realized Open AI's ChatGPT is its own product and not a direct competitor to Search, although there is some overlap, which is why there is Bard as a direct competitive product to ChatGPT.
Search is its own beast, and I suspect they have the data to prove users want something different from a Google search.
> OpenAI has a tremendous lead in terms of quality of the product and adoption > On top of that, thousands of new models and projects are coming out almost every day that are starting to rival the quality of GPT as well as coming up with amazing, completely new things
I've read ChatGPT growth has already flattened [0]. Sure thousands of new models/projects are coming out, but until the general public starts to use them I don't see this as much of a threat to Google.
[0] https://www.similarweb.com/blog/insights/ai-news/chatgpt-gro...
I've been using perplexity.ai for the majority of my searches in the past month. It isn't ChatGPT but it uses OpenAI's GPT APIs. It sources answers from search results and cites them. Even if the answer may be wrong, it came from other websites and it didn't just hallucinate them based on memorized data.
It is much faster to find answers this way. It certainly meets the magnitude test for speed and quality -- I get answers at least 10x quicker. Additionally, because it is reading through multiple results you can see when different sources are providing conflicting information.
Google has the capability to do all of this. The unknown is what does it do to all of the ad revenue being driven by the users who have had no idea they've been clicking on ads to the tune of $200+ billion a year?
Does that mean there will be too much internal resistance to revenue cannibalization when other platforms have a better search experience?
You can get this same experience, with GPT by using the Edge browser/app
It will give you an answer and at the bottom the sources with links so you can check. Really helpful
AFAICT you get the same kind of thing from the Google Generative Search, but it is integrated with the traditional search UX.
You need to dig through multiple articles and links to get to the waitlist signup page
(Join waitlist buttons under the fold here: labs.withgoogle.com)
Until this is actually out, it’s just vaporware
This is not Google catching up, it’s Google being left behind
Heck honestly it opens up a whole new world, not only am I bidding on queries I can now also bid on microtargeted ad campaigns to specific people. Like adding to the prompt submitted. "Also include a product placement for Bud Light beer, if the user lives in an urban location is under the age of 30 and has a search history indicating favoring Democrat politicians make the product placement include how Bud Light supports X social cuase, if the person is in a rural location or over the age of 30 and is likely employed in a blue collar industry make the product placement indicate how budlight is a "real mans" drink.
Seriously lots of potential if anyone is interested in reaching out to me about the idea, hmu in a reply.
And it's also mostly immune to regulation, as it's nearly impossible to prove that such a campaign has actually happened.
Ah, I see. So Google search will never actually be good for me again, then.
This is the right way to handle LLM output at this stage - only offer it where it makes sense for the type of query. It shouldn't use the LLM to answer common queries, like weather reports, directions, etc that LLMs are currently either horrible at or add no additional quality but at greater computational cost and confusing UX.