How does Google possibly know that? Just because they’re shoving them into the top of the search results page, doesn’t mean they’re getting “used.” I have this looming fear that before very long...
AI Overviews are now used by more than 2.5 billion monthly users
How does Google possibly know that? Just because they’re shoving them into the top of the search results page, doesn’t mean they’re getting “used.”
I have this looming fear that before very long the web won’t really be for people anymore. Nobody will click search results because the info they’re looking for doesn’t require visiting another site. Sites will still exist but increasingly their content will be published with agents in mind, not humans. Of course that’s assuming any of them survive the seismic economic shift implied by such a restructuring of how the web is monetized.
Normally I'd agree with you. But I see a lot of people around me use the AI summaries on top for a lot of google queries. Specifically low stakes things where "probably right" is good enough it...
Just because they’re shoving them into the top of the search results page, doesn’t mean they’re getting “used.”
Normally I'd agree with you. But I see a lot of people around me use the AI summaries on top for a lot of google queries. Specifically low stakes things where "probably right" is good enough it seems to be the default for many people now. This includes people in my line of work (software development) but also friends and families. It is very conveniently placed and a lot of people aren't as critical towards AI summaries as you'd hope.
I even had to reprimand myself here as it is such a damn convenience trap to walk into. Combined with the news that search entirely seems to be going away prompted me to properly setup a private SearXNG instance and set it as the default search engine everywhere.
For those not familiar with it, SearXNG is a "metasearch engine" meaning it combines search results from multiple search engines (including google). Added benefits are that with a private engine you can configure a lot of stuff including blacklisting of domains, url rewrites (reddit always goes to old reddit) and much more.
Convenience trap is exactly right. It’s tempting in the same way that crap candy bar is that’s by the till. They have used every psychological technique to try and get you to interact with it.
Convenience trap is exactly right. It’s tempting in the same way that crap candy bar is that’s by the till. They have used every psychological technique to try and get you to interact with it.
I just outright cold turkey'd it by making a unlock origin filter that removes that AI result. Of course, I have no idea if this new rewrite utterly breaks that filter. Very likely will.
I just outright cold turkey'd it by making a unlock origin filter that removes that AI result.
Of course, I have no idea if this new rewrite utterly breaks that filter. Very likely will.
I rarely use Google now, but don't they make you click the button to expand the AI summary? That's a metric they could use to determine if people are using it.
I rarely use Google now, but don't they make you click the button to expand the AI summary? That's a metric they could use to determine if people are using it.
Once upon a time, there was a button that expanded stuff from direct links instead of AI. I still feel like I'm getting tricked into it every time expanding the AI gets me. I'm probably a "monthly...
Once upon a time, there was a button that expanded stuff from direct links instead of AI. I still feel like I'm getting tricked into it every time expanding the AI gets me. I'm probably a "monthly user," but not on purpose.
It does make me wonder what the point of an AI slop page filled with ads and scams would be, if the only things visiting it is bots. My immediate thought is that the economic model they're based...
It does make me wonder what the point of an AI slop page filled with ads and scams would be, if the only things visiting it is bots. My immediate thought is that the economic model they're based on wouldn't work anymore and they might go away. One can hope right?
I'm even finding https://marginalia-search.com to regularly be more useful than ddg or google due to its much more aggressive filtering of spam. My field of search is often heritage building methods and (very broadly) related things, which is obscure, even so, this is impressive/surprising for a one man organisation.
The optimistic view can be that small websites will just be populated by people passionate about things as a hobby instead of some capitalistic ad driven hellscape. The pessimistic view is that...
The optimistic view can be that small websites will just be populated by people passionate about things as a hobby instead of some capitalistic ad driven hellscape.
The pessimistic view is that the single source of truth is just going to become corporate controlled AI that everyone goes to. Not like anyone even goes past the first page of results anyways so what's the difference I guess.
There's tons of metrics which could indicate this. The simplest one, if you make searches but rarely ever open a search result, it stands to reason you're using the above the fold AI overview....
How does Google possibly know that?
There's tons of metrics which could indicate this. The simplest one, if you make searches but rarely ever open a search result, it stands to reason you're using the above the fold AI overview. That signal gets stronger if one clicks to expand the overview, or actually engages with it
Google is practically shoving the AI shit down people's throats, so it would be a lot more surprising to me if people weren't using them. I think the majority of users (my parents, for instance),...
Google is practically shoving the AI shit down people's throats, so it would be a lot more surprising to me if people weren't using them.
I think the majority of users (my parents, for instance), who are not very tech-savvy, don't have a full grasp on how LLMs actually work to understand that the thing they're reading isn't any kind of authoritative answer to whatever question or prompt they just searched for. Since Google's basically presenting it as such, it makes total sense that those masses of people will just take it at face value.
The people who are relatively tech-savvy can fall anywhere on the spectrum between flavor-aid drinking AI tech bro to AI luddite, and at least half of those users will be embracing the AI features, leaving a very small segment of users who wouldn't use them (and some percentage of those users are likely not using Google at all anymore, so they wouldn't factor into the statistics to begin with).
edit: And I guess to actually answer your question, I'm sure Google has statistics around how many users sit at the top of the results where the AI overview is, and interacting with it by clicking the "show more" button or clicking references or asking follow up questions, versus scrolling past it to the actual search results and clicking one of those.
It doesn't exactly matter if they are being deliberately used. The search stopped being about giving users tools to find what they need at around 2015 or so. It is apparently beneficial to the...
It doesn't exactly matter if they are being deliberately used. The search stopped being about giving users tools to find what they need at around 2015 or so.
It is apparently beneficial to the company for them to be used.
Google's AI search tools are a perfect example of "garbage in, garbage out". They suck because their search sucks. Every time I've tried to use regular Gemini after Kagi Assistant, I'm baffled by...
Google's AI search tools are a perfect example of "garbage in, garbage out". They suck because their search sucks. Every time I've tried to use regular Gemini after Kagi Assistant, I'm baffled by how terrible its web searching capabilities are. It also doesn't even cite sources most of the time, so I don't know what it actually found and what it just hallucinated.
They love to show demos of questions like "find me a place nearby that offers this very specific service", so I recently decided to try that and asked Gemini to find a repair shop near me that can do motherboard microsoldering repairs, since I need that. Gemini's response was:
A random repair shop it boldly and in the most bullshitty LLM way ever called "the absolute perfect option". It had no reviews and didn't even do motherboard repairs.
An "alternative option" which was a sketchy repair store with no website and 2 stars on Google Maps, whose "contact" button led to a personal WhatsApp account of some dude
I understand this is not the easiest query. But somehow, Kagi Assistant managed to find me a suitable option first try, and Google is the one who constantly claims to be able to solve tasks like that in all their advertisements.
Kagi Assistant is remarkable. I've been drafting a proposal with Assistant and, in a single thread, the ability to switch between different models was very useful. Any more, I have been using...
Kagi Assistant is remarkable. I've been drafting a proposal with Assistant and, in a single thread, the ability to switch between different models was very useful. Any more, I have been using Minimax for most tasks to keep the cost low, but, when I needed ideas for an explanatory graphic, I switched to the Kagi Research model, which can produce images. Because the whole proposal development process was in the context thread, all I needed to do was ask for the graphic and, voila, there it is. Then I can use Claude Opus for review, once the document is more or less in place. The best thing is that, since Kagi uses enterprise tiers with their providers, none of the queries or responses are used to train the model. It's a brave new world with these tools.
ChatGPT in "thinking" mode does a reasonable job too. I imagine that for Google's AI overview, it's based on the results of just one search, good or bad, while the others do multiple searches,...
ChatGPT in "thinking" mode does a reasonable job too. I imagine that for Google's AI overview, it's based on the results of just one search, good or bad, while the others do multiple searches, refining the query if the first one didn't work very well.
Not entirely surprising. Frankly, I think the UX of natural language conversation is just better than old school search engine querying, particularly the refinement you get as you go back and...
Not entirely surprising. Frankly, I think the UX of natural language conversation is just better than old school search engine querying, particularly the refinement you get as you go back and forth with the LLM. That said, I do concede that it's a bad thing for much of the Internet to get centralized into just a few interfaces and companies, by way of people flocking to LLMs rather than actually visiting the broader Internet. I'm dissuaded from putting any content of worth on the Internet, because it'll just get slurped up into LLMs, and I'll get neither traffic, nor attribution, nor recognition, nor revenue from visitors. Not that I'm against occasional altruism, but for these few centralized traffic hubs to get benefits and profits from my and other people's altruism... well, that's just no good.
That all said, I'll admit my hypocrisy. I use LLMs much more than search engines these days.
Yeah same. Especially with recent updates it’s just so much better. It’s been almost a year since I’ve switched from primarily using Google Search to using Gemini for my queries. Claude and...
Yeah same. Especially with recent updates it’s just so much better. It’s been almost a year since I’ve switched from primarily using Google Search to using Gemini for my queries. Claude and ChatGPT I’ve used for “creative” work and feedback on stuff.
I was super anti-AI two years ago. But it’s hard to continue using stuff that’s not working as well as it used to when there’s an easy alternative.
Have you ever looked closely though at what they search for? I recently have had a close look at the various "deep research" options the big three (Anthropic, OpenAI and Google) offer and while on...
Exemplary
Have you ever looked closely though at what they search for? I recently have had a close look at the various "deep research" options the big three (Anthropic, OpenAI and Google) offer and while on the surface impressive I also found a lot that made me decide against using them for the most part.
To be clear, none of them are exactly transparant about how they perform searching other than the "research plan" the present upfront the and shitload of sources they show you (which, to be frank, seems to be clever UX/marketing to make it look extra legitimate and impressive). Which is already a worrying sign to me, as you can't audit the process itself properly. But even with the lack of information available on the front-end I can still see enough that worries me. A few examples:
"Sources" are taken at face value almost no critical look at the validity of the source, the context it is placed in, etc.
A lot of sources I know are legitimate are rarely included while a lot of listicles, low effort "reviews", etc do make the cut.
In multiple instances when looking closer at the research plan and the "hints" they show during searching it becomes painfully clear that often enough they start with an answer in mind based on training data and try to validate that rather than actually researching the data itself.
Subtly different prompts that by all means should still produce the same factual outcome actually provide wildly different results. This one probably relates to the other points.
In addition to all of this, I also am 100% convinced* that AI powered search is incredibly expensive, more so than traditional search. In my mind this increased cost eventually will need to be paid by someone, which likely is going to be the user. Since the process is non-transparant I am not confident that the results will not end up being polluted by sponsored deals, etc. There is simply no way in my mind that this is going to end up well for us users.
* A while ago I have experimented with creating my own deep research flow with the idea that I might be able to do something with local models. To limit costs I used a SearXNG instance for searching, setup playwright for browsing sources. Using an agentic flow with agents making all the various calls and dispatching other agents ended up eating A LOT of tokens. Even when I did switch to a non agentic flow where each step is orchestrated by code calling on LLMs with simple prompts to validate results still ate a metric ton of tokens for the simplest search query. Mind you, this was not even doing actual deep research but only a few simple search queries. Ironically, google models also did seem to have more trouble coming up with good search queries compared to other models.
Here's a recent reaction which seems to express a sentiment similar to mine (with a perhaps more aggressive tone).
Here's a recent reaction which seems to express a sentiment similar to mine (with a perhaps more aggressive tone).
Your work, your writing or art do matter a bit still: As (unpaid) raw material for their synthetic text extruders. You get to work for free so Google can have tight control
The web is being fully hidden behind a Google-controlled surface.
It’s about monopolizing access to information.
The demos at I/O yesterday seemed cool but I'm not really sure how much value I'd get out of it. Being able to generate apps on the fly is cool in practice but what happens when there's bugs or...
The demos at I/O yesterday seemed cool but I'm not really sure how much value I'd get out of it. Being able to generate apps on the fly is cool in practice but what happens when there's bugs or data? With existing software, there's entire teams dedicated to finding bugs and engineers dedicate part of their time to fix them. But with these, that burden lies on you, the user.
I'm sure the PR response is to ask it to check / fix it's work. In practice with the current models I've messed with that only works so well unless you know the right questions to ask. Google...
I'm sure the PR response is to ask it to check / fix it's work. In practice with the current models I've messed with that only works so well unless you know the right questions to ask. Google seems to be claiming they've fixed that and 'lil timmy can make an OS from scratch just by asking for it, but I guess we'll see.
Kind of torn, because if nothing else there are some abandonware apps ( Minuum keyboard ) that I miss that would be nice to have working recreations of. Who knows how much it would cost or if it could make stable versions that I could trust though.
All this is also an existential threat to the Internet as a whole. If Google isn't driving traffic to websites but just 'doing it for you' a lot of places that rely on ad revenue are going to die off. That has pros and cons but it's definitely a change they're forcing on everyone.
That one struck me as odd anyway. Being able to generate "apps" (simple contained webpages/html files) is something that LLMs have been fairly good at for a while now (I am not sure if you ever...
Being able to generate apps on the fly is
That one struck me as odd anyway. Being able to generate "apps" (simple contained webpages/html files) is something that LLMs have been fairly good at for a while now (I am not sure if you ever saw this, but here is a relevant example as it includes your data from last season of the MC server.
In fact both Anthropic and OpenAI have had this entire thing in their interfaces where it renders in a sandboxed side panel and they have had this for ages. In fact, I thought the Gemini interface has had something similar as well, but either it didn't or this is something new entirely?
Being able to generate apps on the fly is cool in practice but what happens when there's bugs or data?
Depending on the stakes involved that might not really matter. I have no clue how transparant the Google implementation will be but if these are also webapps you can hopefully still validate the source. If not, then I'd agree and only use it for low stakes stuff.
I have very mixed feelings about it. Just yesterday, the HR team sent out their weekly newsletter and this week's topic was some basic best-practices guidelines for using Google Drive. Very...
I have very mixed feelings about it.
Just yesterday, the HR team sent out their weekly newsletter and this week's topic was some basic best-practices guidelines for using Google Drive. Very typical stuff.
Later someone re:all and shared a document they created that explained how they created a tool/app that will create a simple index of everything in a target Google Drive (shared or personal) and save it to a document. They included instructions complete with screenshots, and code to copy paste into Google Apps Script.
On the one hand, I'm very happy that this non technical person was able to create a tool to solve a problem that was important to them.
On the other hand, this feels like Pandoras Box has been flung open. The amount of damage someone could do purely by accident is huge. This is like how mildly technical people learn how to do some basic programming by creating Excel Macros, and end up building spreadsheets that power entire companies. Except way worse.
Well that's terrifying. The feature as such could actualy be very useful assuming it would be created with care, diligence, properly attributed everything, linked to the website first and was left...
Well that's terrifying. The feature as such could actualy be very useful assuming it would be created with care, diligence, properly attributed everything, linked to the website first and was left optional. To be honest it somewhat sounds like simply an llm powered chat integrated into their systems.
Their actual search results were nearly useless for close to a decade now so I do not have cause to expect something different.
Any opinions on where DuckDuckGo is at these days or how it leverages GPT? I find it useful enough for my purposes and when trying Google every so often, the difference is night and day....
Any opinions on where DuckDuckGo is at these days or how it leverages GPT?
I find it useful enough for my purposes and when trying Google every so often, the difference is night and day.
Interesting article. Terrifying, but I'm happy to know what the possibilities are.
How does Google possibly know that? Just because they’re shoving them into the top of the search results page, doesn’t mean they’re getting “used.”
I have this looming fear that before very long the web won’t really be for people anymore. Nobody will click search results because the info they’re looking for doesn’t require visiting another site. Sites will still exist but increasingly their content will be published with agents in mind, not humans. Of course that’s assuming any of them survive the seismic economic shift implied by such a restructuring of how the web is monetized.
Normally I'd agree with you. But I see a lot of people around me use the AI summaries on top for a lot of google queries. Specifically low stakes things where "probably right" is good enough it seems to be the default for many people now. This includes people in my line of work (software development) but also friends and families. It is very conveniently placed and a lot of people aren't as critical towards AI summaries as you'd hope.
I even had to reprimand myself here as it is such a damn convenience trap to walk into. Combined with the news that search entirely seems to be going away prompted me to properly setup a private SearXNG instance and set it as the default search engine everywhere.
For those not familiar with it, SearXNG is a "metasearch engine" meaning it combines search results from multiple search engines (including google). Added benefits are that with a private engine you can configure a lot of stuff including blacklisting of domains, url rewrites (reddit always goes to old reddit) and much more.
Convenience trap is exactly right. It’s tempting in the same way that crap candy bar is that’s by the till. They have used every psychological technique to try and get you to interact with it.
I just outright cold turkey'd it by making a unlock origin filter that removes that AI result.
Of course, I have no idea if this new rewrite utterly breaks that filter. Very likely will.
I rarely use Google now, but don't they make you click the button to expand the AI summary? That's a metric they could use to determine if people are using it.
Once upon a time, there was a button that expanded stuff from direct links instead of AI. I still feel like I'm getting tricked into it every time expanding the AI gets me. I'm probably a "monthly user," but not on purpose.
Yup, happens to me consistently, "oh, fucking God damnit." I usually say.
It does make me wonder what the point of an AI slop page filled with ads and scams would be, if the only things visiting it is bots. My immediate thought is that the economic model they're based on wouldn't work anymore and they might go away. One can hope right?
I'm even finding https://marginalia-search.com to regularly be more useful than ddg or google due to its much more aggressive filtering of spam. My field of search is often heritage building methods and (very broadly) related things, which is obscure, even so, this is impressive/surprising for a one man organisation.
The optimistic view can be that small websites will just be populated by people passionate about things as a hobby instead of some capitalistic ad driven hellscape.
The pessimistic view is that the single source of truth is just going to become corporate controlled AI that everyone goes to. Not like anyone even goes past the first page of results anyways so what's the difference I guess.
There's tons of metrics which could indicate this. The simplest one, if you make searches but rarely ever open a search result, it stands to reason you're using the above the fold AI overview. That signal gets stronger if one clicks to expand the overview, or actually engages with it
Google is practically shoving the AI shit down people's throats, so it would be a lot more surprising to me if people weren't using them.
I think the majority of users (my parents, for instance), who are not very tech-savvy, don't have a full grasp on how LLMs actually work to understand that the thing they're reading isn't any kind of authoritative answer to whatever question or prompt they just searched for. Since Google's basically presenting it as such, it makes total sense that those masses of people will just take it at face value.
The people who are relatively tech-savvy can fall anywhere on the spectrum between flavor-aid drinking AI tech bro to AI luddite, and at least half of those users will be embracing the AI features, leaving a very small segment of users who wouldn't use them (and some percentage of those users are likely not using Google at all anymore, so they wouldn't factor into the statistics to begin with).
edit: And I guess to actually answer your question, I'm sure Google has statistics around how many users sit at the top of the results where the AI overview is, and interacting with it by clicking the "show more" button or clicking references or asking follow up questions, versus scrolling past it to the actual search results and clicking one of those.
It doesn't exactly matter if they are being deliberately used. The search stopped being about giving users tools to find what they need at around 2015 or so.
It is apparently beneficial to the company for them to be used.
Google's AI search tools are a perfect example of "garbage in, garbage out". They suck because their search sucks. Every time I've tried to use regular Gemini after Kagi Assistant, I'm baffled by how terrible its web searching capabilities are. It also doesn't even cite sources most of the time, so I don't know what it actually found and what it just hallucinated.
They love to show demos of questions like "find me a place nearby that offers this very specific service", so I recently decided to try that and asked Gemini to find a repair shop near me that can do motherboard microsoldering repairs, since I need that. Gemini's response was:
I understand this is not the easiest query. But somehow, Kagi Assistant managed to find me a suitable option first try, and Google is the one who constantly claims to be able to solve tasks like that in all their advertisements.
Kagi Assistant is remarkable. I've been drafting a proposal with Assistant and, in a single thread, the ability to switch between different models was very useful. Any more, I have been using Minimax for most tasks to keep the cost low, but, when I needed ideas for an explanatory graphic, I switched to the Kagi Research model, which can produce images. Because the whole proposal development process was in the context thread, all I needed to do was ask for the graphic and, voila, there it is. Then I can use Claude Opus for review, once the document is more or less in place. The best thing is that, since Kagi uses enterprise tiers with their providers, none of the queries or responses are used to train the model. It's a brave new world with these tools.
ChatGPT in "thinking" mode does a reasonable job too. I imagine that for Google's AI overview, it's based on the results of just one search, good or bad, while the others do multiple searches, refining the query if the first one didn't work very well.
Not entirely surprising. Frankly, I think the UX of natural language conversation is just better than old school search engine querying, particularly the refinement you get as you go back and forth with the LLM. That said, I do concede that it's a bad thing for much of the Internet to get centralized into just a few interfaces and companies, by way of people flocking to LLMs rather than actually visiting the broader Internet. I'm dissuaded from putting any content of worth on the Internet, because it'll just get slurped up into LLMs, and I'll get neither traffic, nor attribution, nor recognition, nor revenue from visitors. Not that I'm against occasional altruism, but for these few centralized traffic hubs to get benefits and profits from my and other people's altruism... well, that's just no good.
That all said, I'll admit my hypocrisy. I use LLMs much more than search engines these days.
Yeah same. Especially with recent updates it’s just so much better. It’s been almost a year since I’ve switched from primarily using Google Search to using Gemini for my queries. Claude and ChatGPT I’ve used for “creative” work and feedback on stuff.
I was super anti-AI two years ago. But it’s hard to continue using stuff that’s not working as well as it used to when there’s an easy alternative.
Have you ever looked closely though at what they search for? I recently have had a close look at the various "deep research" options the big three (Anthropic, OpenAI and Google) offer and while on the surface impressive I also found a lot that made me decide against using them for the most part.
To be clear, none of them are exactly transparant about how they perform searching other than the "research plan" the present upfront the and shitload of sources they show you (which, to be frank, seems to be clever UX/marketing to make it look extra legitimate and impressive). Which is already a worrying sign to me, as you can't audit the process itself properly. But even with the lack of information available on the front-end I can still see enough that worries me. A few examples:
In addition to all of this, I also am 100% convinced* that AI powered search is incredibly expensive, more so than traditional search. In my mind this increased cost eventually will need to be paid by someone, which likely is going to be the user. Since the process is non-transparant I am not confident that the results will not end up being polluted by sponsored deals, etc. There is simply no way in my mind that this is going to end up well for us users.
* A while ago I have experimented with creating my own deep research flow with the idea that I might be able to do something with local models. To limit costs I used a SearXNG instance for searching, setup playwright for browsing sources. Using an agentic flow with agents making all the various calls and dispatching other agents ended up eating A LOT of tokens. Even when I did switch to a non agentic flow where each step is orchestrated by code calling on LLMs with simple prompts to validate results still ate a metric ton of tokens for the simplest search query. Mind you, this was not even doing actual deep research but only a few simple search queries. Ironically, google models also did seem to have more trouble coming up with good search queries compared to other models.
Here's a recent reaction which seems to express a sentiment similar to mine (with a perhaps more aggressive tone).
The demos at I/O yesterday seemed cool but I'm not really sure how much value I'd get out of it. Being able to generate apps on the fly is cool in practice but what happens when there's bugs or data? With existing software, there's entire teams dedicated to finding bugs and engineers dedicate part of their time to fix them. But with these, that burden lies on you, the user.
I don't think they're real apps in the way we typically think of them. It's more like some features running in the Gemini sandbox.
I think of them more like web apps I guess. Instead of the browser being the platform, Gemini's sandbox is the platform.
I'm sure the PR response is to ask it to check / fix it's work. In practice with the current models I've messed with that only works so well unless you know the right questions to ask. Google seems to be claiming they've fixed that and 'lil timmy can make an OS from scratch just by asking for it, but I guess we'll see.
Kind of torn, because if nothing else there are some abandonware apps ( Minuum keyboard ) that I miss that would be nice to have working recreations of. Who knows how much it would cost or if it could make stable versions that I could trust though.
All this is also an existential threat to the Internet as a whole. If Google isn't driving traffic to websites but just 'doing it for you' a lot of places that rely on ad revenue are going to die off. That has pros and cons but it's definitely a change they're forcing on everyone.
That one struck me as odd anyway. Being able to generate "apps" (simple contained webpages/html files) is something that LLMs have been fairly good at for a while now (I am not sure if you ever saw this, but here is a relevant example as it includes your data from last season of the MC server.
In fact both Anthropic and OpenAI have had this entire thing in their interfaces where it renders in a sandboxed side panel and they have had this for ages. In fact, I thought the Gemini interface has had something similar as well, but either it didn't or this is something new entirely?
Depending on the stakes involved that might not really matter. I have no clue how transparant the Google implementation will be but if these are also webapps you can hopefully still validate the source. If not, then I'd agree and only use it for low stakes stuff.
I have very mixed feelings about it.
Just yesterday, the HR team sent out their weekly newsletter and this week's topic was some basic best-practices guidelines for using Google Drive. Very typical stuff.
Later someone re:all and shared a document they created that explained how they created a tool/app that will create a simple index of everything in a target Google Drive (shared or personal) and save it to a document. They included instructions complete with screenshots, and code to copy paste into Google Apps Script.
On the one hand, I'm very happy that this non technical person was able to create a tool to solve a problem that was important to them.
On the other hand, this feels like Pandoras Box has been flung open. The amount of damage someone could do purely by accident is huge. This is like how mildly technical people learn how to do some basic programming by creating Excel Macros, and end up building spreadsheets that power entire companies. Except way worse.
Well that's terrifying. The feature as such could actualy be very useful assuming it would be created with care, diligence, properly attributed everything, linked to the website first and was left optional. To be honest it somewhat sounds like simply an llm powered chat integrated into their systems.
Their actual search results were nearly useless for close to a decade now so I do not have cause to expect something different.
Any opinions on where DuckDuckGo is at these days or how it leverages GPT?
I find it useful enough for my purposes and when trying Google every so often, the difference is night and day.
Interesting article. Terrifying, but I'm happy to know what the possibilities are.