And yet, ironically, we have this post. And sites like Reddit and Substack and the site I'm commenting on right now -- lots of human thought still in circulation on the Internet. The sites he mentioned like Quora are dying for other reasons. And LLMs still pull from all these sources, perhaps more draining than search engines that came before LLMs, but yet we are still here, posting and commenting, I don't see it going away. In fact, I meet more and more friends that are finding their voice on Substack or YouTube or their own blog and producing even more thought content on the web than ever before.
I think reddit is kind of the poster child of sites being hollowed out by Ai posts. The vast majority of posts nowadays are AI reposts responded to by AI comments, usually trying to push some marketing or political drivel.
This project is awesome. I'm thinking of mounting it on an fpv quadcopter to get a realtime survey of the RF situation below. What is the feasible range triangulating where RF activity is located below in a situation like that? If it can work at 50ft range I could make it work but I'm wondering if that is realistic with this kit.
A lot of his ideas remind me of the BeReal app, it limits posts per day and is geared toward 'friends in real life' and with just a few friends on it I've stayed engaged. But it's sparse for me and can be a ghost town much of the time, but that may be just because my friend group isn't using it much. There needs to be sufficient network effect to maintain and grow it's reach as a network, which may be antithetical to its founding principles.
Lol I dont know how to say this but BeReal is flawed from the start. IG is not about being real at all for the most part - hence its popularity. People enjoy being able to cosplay and show their best self, not revealing their true self.
That's been my experience with BeReal as well. I seldom see posts from people I'm close with, and so I post less as well, and presumably that contributes to a similar feeling for other friends of mine.
In fact, these days, I only post in it so that I can record the moment, to add it to the record of fotos which are convenient and fun to look back through.
I liked BeReal while it was still just post 1 picture a day. Now you have 'your brands' there, they try to increase 'engagement' by trying to get people use the app more... I was there when it happened on Facebook, I was there when it started on Instagram – not again, thx.
They show the factory in their video on the page, it's a new modern car chassis with the vintage look (so it's not a Ford Bronco chassis but the external look is inspired by it).
Sadly the past link is a little more basic - just the title, or the URL (if it's constructed reasonably). It doesn't find all related posts (which is still NP)
This action comes a bit late, at the end of the "Search engine" era, at a time when AI responses from many sources are largely replacing the "Google Search".
Similar action happened against Microsoft Windows around 2000, just as the rise of web-based apps (online email, google docs, etc) largely made the underlying operating system less relevant to how people use their computers and apps.
So I read this as the dominant player can monopolize a market while it's relevant without an issue, and once the market starts to move on, the antitrust lawsuits come in to "make the market more competitive" at a time when that era is mostly over.
And trying to regulate early (as with the last administration's AI legislation that is now being repealed) we can see that only hindsight is 20/20, and regulating too early can kill a market. My conclusion is to just let the best product win, and even dominate for a while as this is part of the market cycle, and when a better product/platform comes along the market will move to it.
Because we're still in the "get them hooked" stage where AI startups and Google/Meta are losing money on AI.
Once they start properly monetising (aka making users pay for what it actually costs them to train and run), it will be a different story. The vast majority of people won't pay $20-30€/month for an LLM to replace their search engine. (And an analysis I saw of OpenAI's business model and financial indicated they're losing money even on their paid tier, per query).
They’ll instead accept being manipulated and fed content influenced by money, be it ads or individualized content that somehow serves whoever is paying for it.
IMO the AI players are only hooking the power users and technically inclined (in regard to AI replacing search), but there's a long, long tail of people who are going to be using Google as their search engine until the end of days. Think the type of person who types "facebook.com" into Google instead of their address bar — they're not switching over to ChatGPT or Kagi any time soon.
I’ve started using ChatGPT to look up most things I used to Google. It gives me immediate, concise results without having to parse bloated websites full of ads and other garbage.
Google themselves are trying to figure this out, with the first (top placement) of search results showing their Gemini AI Response, at least for me. I read this as an attempt to keep users on Google instead of asking Chat GPT or other some other AI. What's your take on that?
My take is that it's primarily a smart way to quickly gather lots of ELO/AB feedback about LLM responses for training, whilst also reducing people switching to ChatGPT. OpenAI has significant first mover advantage here and it's why they're so worried about distillation, becuase it threatens the moat.
Google, on the other hand, has a huge moat in access to probably the best search index available and existing infrastructure built around it. Not to mention the integration with Workspace emails, docs, photos - all sorts of things that can be used for training. But what they (presumably) lack is the feedback-derived data that OpenAI has had from the start.
ChatGPT does not use search grounding by default and the issues there are obvious. Both Gemini and ChatGPT make similar errors even with grounding but you would expect that to get better over time. It's an open research question as to what knowledge should be innate (in the weights) and what should be "queryable", but I do think the future will be an improved version of "intelligence" + "data lookup".
Every AI chatbot to date suffers from the "Film expert" effect. That is when a script writer presents data from an "expert" in a movie or show to the audience in response to some information need on the part of the other characters. Writers are really good at making it sound credible. When an audience member experiences this interchange, generally they have one of two experiences. Either they know nothing about the subject (or the subject is made up like warp drive nacelle engineering) and they nod along at the response and factor it into their understanding of the story being told. Or, they do know a lot about the subject and the glaring inaccuracies jolt them out of the story temporarily as the suspension of disbelief is damaged.
LLMs write in an authoritative way because that is how the material they have been trained on writes. Because there is no "there" there, an LLM has no way of evaluating the accuracy of the answer it just generated. People who "search" using an LLM in order to get information about a topic have a better than even chance of getting something that is completely false, and if they don't have the foundation to recognize its false may incorporate that false information into their world view. That becomes a major embarrassment later when they regurgitate that information, or act on that information in some way, that comes back to bite them.
Gemini has many examples of things it has presented authoritatively that are stochastic noise. The current fun game is to ask "Why do people say ..." and create some stupid thing like "Too many cats spoil the soup." That generates an authoritative sounding answer from Gemini that is just stupid. Gemini can't say "I don't know, I've never seen anything that says people say that."
As companies push these things out into more and places, more and more people will get the experience of believing something false because some LLM told them it was true. And like that friend of yours who always has an answer for every question you ask, but you keep finding out a bunch of them are just bullshit, you will rely on that information less and less. And eventually, like your buddy with all the answers, you stop asking them questions you actually want the answer too.
I'm not down on "LLMs" per se, but I do not believe there is any evidence that they can be relied on for anything. The only thing I have seen, so far, that they can do well is help someone struggling with a blank page get started. That's because more people than not struggle with starting from a blank page but have no trouble correcting something that is wrong, or re-writing it into something.
"Search" is multifaceted. Blekko found a great use case for reference librarians. They would have paid Blekko to provide them an index of primary sources that they could use. The other great use is shopping if you can pair it with your social network. (Something Blekko suggested to Facebook but Zuck was really blind to that one) Blekko had a demo where you could say, "Audi car dealer" and it would give you the results ranked by your friend's ratings on their service. I spent a lot of time at Blekko denying access to the index by criminals who were searching for vulnerable WordPress plugins or e-commerce shopping carts. Chat GPT is never going to give you a list of all sites on the Internet running a vulnerable version of Wordpress :-).
So my take is the LLM isn't a replacement for search and efforts to make it so will stagnate and die leaving "Search Classic" to take up the slack.
If you trained a model on a well vetted corpus and gave it the tools to say it didn't know, I could see it as being a better "textbook" then a physical textbook. But it still needs to know what it doesn't know.
What happens when search engines start displaying AI generated/human seasoned content? Everything written online is being seasoned by ChatGPT et al. and human behavior is being conditioned to think by its highly convincing outputs.
Why should I trust an ad supported search engine with the bastardized search experience displaying top rank results by highest paying advertiser and now mixing in AI slop pretending to be human written for the foreseeable future. The future looks bleak..
I rather create an AI agent to do my Google search for me and cut out the ad links results bias and further synthesize the results for me in human readable and interrogabal format.
I will take that bet Chuck. Maybe not completely replace, but AI will be the defacto search platform. I find myself using Google less and less these days
I only use Google when I am in the mood to search for something and find it below the fold after scrolling through mindless semi-related sponsored links
> This action comes a bit late, at the end of the "Search engine" era, at a time when AI responses from many sources are largely replacing the "Google Search".
AIs are using search indices more and more. Google has the largest, and there is risk of Google using its monopoly in search (in particular their index) to give themselves an unfair advantage in the nascent AI market.
Do you think its a coincidence that companies start getting regulated when they are at the end of their rope? I think its the other way around. When Google started to lose power, when alternatives started to pour money into the political system, when jobs and money from alternatives presented themselves, then the wheels of justice began turning. The regulation is caused by googles apparent decline.
"Nuclear made up close to 5% of China's power generation in 2024 and is expected to rise to 10% by 2040". That is the stat that surprised me. On balance, I understand that nuclear is environmentally much cleaner than coal, correct? If 95% of power in China is coal (with some hydro/solar/wind) then perhaps the more nuclear, the better.
I have not used this one yet but as a rule of thumb I always test this type of software in a VM, for example I have an Ubuntu Linux desktop running in Virtualbox on my mac to install and test stuff like this, which set up to be isolated and much less likely to have access to my primary Mac OS environment.
I'm just in the process of setting this up using logs from a tree that just went down in a storm, so thanks for pointing that out, I had not thought of it yet! Last year I did have to build a cage around our strawberry patch and it makes sense that we will need one over this project as well. Much appreciated.
