The way I see it:<p>- Today's search engines will give you links to millions of documents<p>- ChatGPT, if asked correctly, will instead generate one perfect document based on millions of the documents<p>To me, that's a clear evolution of the search engine, especially with all the SEO & ad spam that's plaguing Google and others currently.<p>I wasn't ready to pay a monthly fee for an ad-free Google. I am ready to pay for something like ChatGPT.<p>Google has an issue on their hands and is probably working overtime to lobby the threat of ChatGPT away.
I needed to write an official letter today.<p>I could have written it myself, but that would take some effort.<p>I could google it, scroll past the ads. Click on the first result, hope it is kinda what I need. On that page scroll past the ads and SEO blabla-text. Copy the information that I need. Paste it in a doc. Rewrite it to what I need. Fill in my information.<p>I went to ChatGPT told him what kind of letter I need. Told him what personal information it should put in. Got a good enough text back. Told it what to correct. Copy pasted it and formatted it a bit. Done.<p>way better experience then with Google.<p>Google is in trouble, and they know it.
They are not a replacement, but unfortunately search engines are turning into AI chatbots too.<p>When I think of a search engine I want a "grep for the Internet", not an "AI".<p>Looking up part numbers for ICs and other electronic components is the most prominent application where search engines like Google have gotten far worse in recent years, and AI ain't going to work there either.
I think AI chatbots will replace some of the use cases for search engines, but I also think that is a good thing. It will let search engines do what they do even better. Search engines right now do a lot of things, arguably a bit too many, which has the actual search-aspect suffer a bit.<p>A significant design obstacle with almost all search engines is that it's not entirely clear what you can do with them. They're just a search box, after all. You need either external training or a trial-and-error loop to figure out what you can do with them. This also means that if you change how they work, users get confused because their queries no longer give them what they expect (this is likely a big contributor to the "google is broken"-sentiment that's been going around).<p>This becomes a problem when the search engine caters to too many modalities of search, as there's an element of guesswork what the user want. Two users may even enter the same query, but look for different results. If I enter the name of a video game, do I want to buy it, or do I want information about it, or maybe fanfiction or something other than the game that shares the same name? Who the heck knows.
I agree that chat bots aren’t the proper modality for replacing search. So what? Currently, Google search results stink. ChatGPT results are way better in a number of domains. Does it need to be a chat bot? No. But Google still stinks now. I’ll take anything that can just find the correct information.
The big drawback chatGPT has is that on many topics it walks on eggshells.<p>It can’t give me a direct answer. It couches the answers in nonsensical caveats. Adding stilted context that really does not add value to an answer and actually makes the search more tedious.<p>If I ask it the male female breakdown for crime statistics it begins to get defensive and gives me general answers. I can prod it to finally give me government statistics but it doest it begrudgingly. And that’s for a far away country not steeped in any unusual crime controversy.
Today I wanted to mute the bark on windows 11.<p>Not knowing the terminology, Google was useless.<p>I ask chatgpt and it tells me how to mute notifications, I clarify "no the thing where you're typing and windows can't do it so it plays a sound" and chatgpt happily informed me on how to mute the windows hard stop sound.<p>For situations where you don't really know what you're looking for, chatgpt is already competitive with Google. Failing abysmally in some cases, and far surpassing Google in others.
For me, ChatGPT has already replaced Google for most queries, so the point is moot.<p>Even if ChatGPT is flawed, it's still way better than Google. Using Google feels stupid; trying different search terms and clicking through dozens of websites with cookie popups and skimming all the bloated content for an answer which might be close to something you're looking for. ChatGPT gives the answer instantly.<p>I don't believe there's any future for Google (and Google-optimized websites).
It seems the only thing that's missing is some type of fact-checking function. The interaction, from a user perspective, is much nicer than sorting through Google results.But the results can be confidentially wrong and if you're not familiar with the subject matter already, you won't really know that.<p>That said, I'm basically using it as a replacement for Google for stuff that isn't up-to-date (code, philosophy) then double checking the output to see how it's wrong.
Google is more "chatty" than it was a few years ago and ChatGPT is a quickly moving target it seems - it's answers seem more "search-like" than they seemed when I started playing with it just a few weeks ago (more caveats and more likely to give multiple options, etc). It seems like we'll have fusion soon.<p>I agree the results will be unpleasant. I already despise Google's fucking "looks like there aren't many results" message and there will be more to hate down the road. But still, appearing to give "an answer" rather than reporting information seems like a winning quality to bring in the masses. As someone pointed out, Google's target audience is inherently those credulous enough to be valuable targets for their advertisers.
Not sure if I'm missing something but if the authors claim:<p>1. That chatbots don't understand what they are producing -> a search engine doesn't either.<p>2. The benefit of search is that you engage in sense-making -> you actually need to do that with a chatbot too.<p>3. You would want access to the sources -> that's already fixed for today's chatbots (see perplexity.ai, for example, which offers links next to its claims).<p>4. We may trust chatbots more easily because of their language: we already have that problem with search results.<p>I understand the idea that there will not be an ever-knowing AI and we will probably don't even want one. But what we have today seems an improvement over a search engine (not necessarily a replacement).
Eventually Chatbots will start inserting product placement ads as well. So comparing the ads on Google with the ad free ChatGPT experience is detracting from the real value of ChatGPT.<p>Google results are mostly ok. But I have to do the synthesis. ChatGPT does the synthesis for me saving me time and mental bandwidth. This is the part that I find valuable.
I mean, ChatGPT came out 2 weeks ago.<p>By the time Google even was started, Excite had been searching the web for 7 years and Lycos had been around for 3 years.<p>I don't think it's a 1:1 replacement, but let's not judge a product that's been out for less than a month against a trillion dollar behemoth that launched 24 years ago.
I had a recent example of ChatGPT clearly beating Google:
"how to debug a Chrome extension in VSCode?"<p>After 10 minutes of reformulating this, I couldn't get a good answer from Google, but ChatGPT got it on the first try. Many Google results were about a depreciated extension for VSCode called Chrome Debugger. Maybe if I was normally doing a lot of front-end coding, it would be trivial.<p>Google needs to get going with their version of fine-tuning PaLM from human preferences, get a good classifier based on user characteristics and query to decide when to use it, some checks to make sure they're not committing copyright infringement, and some text ads in the middle of the generated text, and mark it experimental and let it loose on the search results pages. Listing references, online updates, etc. can be added later. It won't be profitable even if they only enable it on a small percentage of queries, but they're risking the whole business if they don't do it.
What the near future would look like is-
1. Google: will implement some form of LLM on their engine and the contextual answers will significantly improve(~close to what ChatGPT produces) satisfying most people. But also displays the top links that contributed to the context. This is difficult to do but somehow they gotta figure out or risk their entire core business getting affected.<p>2. ChatGPT and other LLMs: will take some more time to meaningfully compete with google(as in becoming a noun) and in the mean time they would provide APIs to their models so that large orgs(businesses, scientists etc) can plug it in with their own data and use. They may or may not charge a fee for it.<p>In both cases there is one big pro and con:
Pro: Death of SEO and ads as we know it
Con: The model dictates what majority of internet users read, change their views or manipulate. And eventually becomes an AI weapon.
Search engines are and anomaly. They started as indexes of content and have remained in that form.<p>I still have to do all the work of synthesising the knowledge I need from the results.<p>It’s like having to go through the list of references of a scientific paper or Wikipedia article without having access to the article itself.<p>ChatGPT is giving you the article, it summarised that knowledge for you, a much effective way of getting what you actually want.<p>The ideal would be a combined system where I get my answer and also the relevant references that shaped it.
Today I experienced just the opposite. Chatgpt answered some business questions in less than a minute. I would have spent 30 mins in seo optimized sites to find the exact same info.
I worked on a chat bot to help solve technical issues. It would parse what the user wanted and then search the already established articles we had that were ingested into Elasticsearch.<p>At first, we started with a goal of fully conversational AI. So, for example, it would ask a question based on the article and give you a choice, you could then type your choice. This became a nightmare for the model so we added buttons instead.<p>Then, before we knew it, this “bot” was just a glorified search engine that feigned being a bot. Towards the end of it I scratched my head and said “did we even replace the current knowledgebase? are we spending thousands on something that adds no value?”<p>I don’t think chatbots are it either. I think we could have just replaced the KB search with Elasticsearch and been done with it, no need for any ML.
This was written by a human? It could have been but it reads like it was unfolded from 1 perhaps 2 lines of.. lets say beef. There is very little going on in the text, the blog post is the [now usual] short form free from interesting references to previous writings around this argument, as if the topic was entirely original and <i>invented here</i><p>Good references, more than anything, suggest to me an author knows his topic.<p>The www as a festival of sharing knowledge is really not what we thought it would become or perhaps could or should have.<p>The author, like most, is definitely capable of writing, sourcing and building(!) something into the chronology of the topic.<p>But ok, to entertain the idea: I think a chat bot the way companies and government tried to glue it onto their knowledge base (and mostly failed) will now become a real possibility, it will be able to find the most likely document you are looking for and know when the match is to crappy to mention.<p>An all knowing AI is not required. Have one per domain and let them exchange words.<p>Real time learning from the conversation is still off but I imagine that could easily change.<p>Thinking about it, if you want request all information about you from the data hoarders you could build quite the interface to represent you in the bot net. It doesn't have to know hoe to fill taxes as long as there is a tax bot available for 7.3241 dollar per query. Your personal assistant can learn the bits of tax code relevant to your disposition. Seems fun.
Yeah, but they will be. I use it exclusively instead of Google for fact finding. Yes, you can't find a restaurant with it. But it's incredible for teaching programming and general known knowledge.<p>I cannot understand why the tech crowd has blinders on for AI. I'm 40 years old and have been a programmer since I started playing with BASIC since I was in elementary school.<p>The shit GPT is doing, I thought I'd never seen in my lifetime.<p>Maybe people are afraid of being replaced, or not being "special magical beings". Y'all are biological robots. You need to accept that.<p>AI is going to trounce humans in literally everything in a matter of a few years.<p>Maybe it's an existential problem, I don't know, but I feel everyone is missing the forest for the trees.
They are exactly that - search engine evolution. I don't want to search through 100s of pages looking for an information - I want information based on the context I provided and then references to relevant pages if I need them. If you provide that no one will want to use regular search anymore. I don't think that we are there yet, even with next gen. gpt4. But it is inevitable to happen.
My prediction is - if Google doesn't release competitor to ChatGPT in 2023 it is doomed in the next 5-6y, just like Yahoo. And Microsoft finally wins search engine war.
This is a highly editorialised title. The title chosen by the authors is "All-Knowing Machines Are A Fantasy". There is nothing in the blog post about "AI chatbots" versus search egnines.<p>Due the editorialised HN title the top comment takes the thread on a tangent about ChatGPT versus Google web search.1 There is nothing in the article about Google specifically, other than to mention Google has their own language model for dialogue ("AI chatbot") along with OpenAI and Facebook. Nor is there any discussion of search engines, e.g., Google's, except to point out at the end that "AI" and search engines are similar because neither is under the control of the user. They are each offered by corporations that, regardless of the input, shape the output to benefit the corporation's own commercial interests. Neither offers "neutral" results.<p>A true improvement in search technology would be one that provides neutral results. Providing that improvement means the "tech" company cannot survive. It has no viable business model except operating as an intermediary (middleman), selling ad services.<p>1. Incidentally no user ever verifies that either Google or ChatGPT searches through millions of anything. The corporation will tweak the results to make sure it stays in business and maximises profit. The user is never allowed to see the tweaks.
The way I see it is that Google (search engines) is a particular solution to a set of retrieval information problems. ChatGPT does seem to be a good solution for a different set of problems. As others have pointed out both have their benefits and limitations.<p>The future of how we retrieve knowledge might use a combination of these technologies along with others that may not yet exist.<p>ChatGPT is definitely showcasing how good the future systems will be at understanding natural language input, as well as producing answers in the appropriate form.
AI chatbots(or equivalent AI answering machines) vs Search Engines<p>Search has 2 kinds of users.<p>1 trawls through multiple pages of results to find what they are looking for. This is you and me but we are in minority.
2 clicks on the first link(or sometimes 2nd) and that's the end of that particular search. The majority.<p>2nd will decide the winners<p>Academics, being the 1st kind are the reasons for arguments like those in the article. They are not wrong but to the majority, it won't matter.<p>And I am willing to guess the Code Red at Google is about the 2nd type of users. They are the bread and butter.
Q: how might an AI algorithm be modified in order to return citations with its response?<p>A: There are several ways in which an AI algorithm could be modified to return citations with its responses. Here are a few possibilities:<p><pre><code> One approach would be to add a citation generation module to the AI algorithm. This module could be trained on a dataset of citations and could be used to generate appropriate citations for the information provided by the AI algorithm.
Another option would be to modify the AI algorithm to include a database of references, with each reference being associated with a specific piece of information. When the AI algorithm provides a response, it could also return the appropriate reference(s) from the database.
A third possibility would be to modify the AI algorithm to search external sources of information (such as online databases or libraries) for relevant citations whenever it provides a response. The AI could then return the relevant citations along with its response.
</code></pre>
It is worth noting that adding a citation generation module or incorporating external sources of information into an AI algorithm can be a complex task, as it requires the development of additional algorithms and data structures to support these features. It may also require the acquisition of additional data or resources to support the citation generation process.
I used ChatGPT last week to accelerate a “DevOps” project that I had. It involved writing a bunch of relatively simple Python scripts that transformed source JSON to specifically formatted yaml.<p>I just told it my input and my desired output and it wrote the code perfectly. Then I kept refining the script with more specific instructions.<p>I also needed to document sone scripts that I previously wrote. I just copy and pasted the script into ChatGPT and asked it what does it do and used that for my README.<p>I even used it to create a simple FAQ for my scripts and asked “Why would you use it”? The answer to why you would use my scripts jibed with the reality.<p>Another use case was that I was trying to get a friend interview ready for an eventual AWS tech support position. I just sent him a series of interview questions that I would ask at an interview and told him to put those questions in ChatGPT (I had already vetted a few).<p>I would ask questions like “compare and contrast X and Y”? “when would you use X”? “how can you save money with Y”?
The question is not whether AI will generate the right answers, but whether it will generate better answers than a search engine. Given that search results are dominated by ad-supported websites, maybe an AI results, which are 99% accurate, will be better than whoever spend the most on ad words.
It's 2045, the Sino-AI war has devastated the climate and huge dust clouds cover the earth. Human-AI hybrids are forming a new government. A new deadlier virus engineered in a lab is set to wipe out humanity. People are still adding "reddit" to their search terms.
All of these criticisms fail to address a big elephantine question: Are these hurdles insurmountable?<p>If ChatGPT's model fails to differentiate or understand the question of credibility, how tough would it be to build classifiers that could do a 75% decent job?<p>If AI presentation of knowledge is imperfect - well, fine. But is human presentation or selection perfect for that matter? Criticising a thing for imperfection is not a viable take.<p>If ChatGPT's knowledge is limited, how tough would it be to plug it into larger and larger datasets, including the live internet itself?<p>It's like there's a wall between the people who see the potential of this and the people who feel so threatened by it that they refuse to look over the wall.
When ChatGPT came out I used it to do 100 hours worth of work in 10 hours or so. But it still felt like work. And for more intricate use cases it is just easier to do traditional research.<p>We’re safe for a while still. And when I say we I mean writers.
Recent paper showing how LMs "struggle with less
popular factual knowledge, and that scaling
fails to appreciably improve memorization of
factual knowledge in the tail". In other words, simply scaling LLMs will not result in replacing Google as a search engine. Approaches like theirs ie retrieval augmentation are needed.<p><a href="https://akariasai.github.io/files/llm_memorization.pdf" rel="nofollow">https://akariasai.github.io/files/llm_memorization.pdf</a><p>This is essentially what I'm building for the content marketing vertical.
It's like putting a dusty book back on the digital shelve, now you need to have a conversation to do a simple search query that's hidden in marketing gibberish?<p>I'm, still seeing the whole push in this product range hang with the call to make the search functions of most search engines more garbage than the actual chatbot.<p>As if Marketing is steering this ship.<p>I'm sticking to my own knowledge and trusted independent sources for now, no need to have something make up an entire world for you, just to trap you in a cage of ones own making for real this time?
maybe not search engines, but it has totally replaced stackoverflow and google for searching for docs or even wikipedia in some sense (e.g. make alpha blend between two argb uint32 pixels)<p>google has been going progressively worse to find what i need, sometimes i even have to go to second page, wasting time clicking on bullshit markov chain seo sites to find out how to parse date in whatever new language i am learning<p>with chatgpt: parse a date from string in the v programming language, try googling that
I feel like most people are trying to give personality to AI chatbots. From what I have read, ChatGPT is good at carrying conversations and doing text based work. I would focus on feeding it curated info and making its answers as factually correct as possible. And some way to know the sources of information it picked up answers from. Companies and people are going to do that anyway in the coming future. We may see chatbots good at a certain field, and whose results can be trusted.
I compare it all to using Github Copilot in my daily work. Before, I needed to Google many problems to find answers, curate the answers, scroll down websites, read downvoted comments on StackOverflow, copy/paste code, refactor it, turn code into TypeScript, etc.<p>Now, I just tell Copilot what I need to do, and I simply curate multiple solutions.<p>ChatGPT also understands what I need to do and changes things based on human instructions. It gets things right much more quickly.
An intriguing headline with very little actual substance. Nothing is a direct replacement for anything if we're pedantic enough.<p>Disclaimer: I stopped reading at the paywall.
AI chatbots are (or will be) a replacement for the user interface of search engines.<p>They'll still be backed by search engines.<p>(And this isn't news; Google has been saying this for well over a decade, I think, by this point, and Google, Amazon, and others have been <i>doing it</i> with simpler chatbots, primarily for voice assistants, for some time; LLMs are a big step forward for this—though I suspect that per request resource usage is <i>far</i> more.)
ChatGPT has its uses, but it can be dangerous: The other day it told me that the legal limit for Plutonium in drinking water is 0.1 micrograms per liter.
Awful article, probably has a lot of upvotes because people agree with the title.<p>> But the systems do not have any understanding of what they are producing, any communicative intent, any model of the world, or any ability to be accountable for the truth of what they are saying.<p>Same is true for search engines. (LLMs could include links to the relevant source material too, if that counts as accountability.)
They are not, however the search engine function will now be more specific. Do you want to discover sources or do you just need information?<p>Keep in mind that AIs (have already) will infest the search engines by creating websites with the content you search ad hoc and many other tricks. If there is money in it, AIs will take over it, ultimately for the benefit of the user.
The article does not show up without JS, so I'll answer the title.<p>I agree. In practice, you'll likely need to double check what it says using a search engine anyway. As long as the Chatbot does not provide the sources for whatever it states, I won't trust it. There's no reason to. Unless it produces pure logic that can be checked using reasoning.<p>… as I have in real life from people unless they are expert in the area / can be trusted to varying degrees for some reason, or from anything on the internet, including Wikipedia. AI-based stuff hasn't changed this fundamental principle.<p>ChatGPT is trained on Wikipedia (which is mostly fine, though you still need to check the sources!), plus random things where people could have stated random stuff, maybe jokingly, or could have been sarcastic or whatever. So the training set is already not trusty. And even on a trusty set, the process is so opaque that we don't know what plausible but false rehash it's capable of outputting. Actually we know that it produces what has been called "hallucinations", so…<p>It's easy to get lazy about double checking the overly confident tone or the flawless writing of the chatbot. It's probably a good way to explore, but it can't be fully trusted yet. Even for exploring, I'm afraid of risking being influenced by reading a false piece of information written perfectly, even if I know I should double-check it.<p>However, the day the chatbot provides the sources, it might change a lot of things.
Right now I find ChatGTP excellent for certain technical topics I'm familiar with, but can't recall the details. It saves me the time of looking for an example from stackoverflow/tutorial blogs/official documentation, and gives me just the piece I want - things like syntax, which libraries/modules to use etc.
I think this article misses the point. I don’t imagine ChatGPT as is to be a replacement for a search engine. I do however see the interface of talking to an AI in a REPL sort of way, where the context is carried through the conversation as a massive improvement with tons of potential that is yet to be realized.
Not yet 1:1 replacement, but AI chatbot agents that are capable of interpretable chain of thought reasoning and conduct searches to fill gaps in knowledge and check their sources will be highly disruptive to legacy search and will supplement, and even replace legacy search engines in many use cases.
An important point about the authors of this post. They are very biased (at least Bender) because they are academic who have been fighting large language models from industry for years now. The main reason is that they make all their research obsolete so they fight to keep their relevancy.
It is incredible that we as humanity are about to put our info search in hands of a hallucinatory text generator, without certainty about (i) how resilient the whole thing is against spam/propaganda, and (ii) what happens to people unable to find the difference.
No shit, people are replacing product support with it, not search engines...<p>Also OP fucked up re-redacting title from sense into nonsense.<p>But it would be interesting seeing ChatGPT-like but learned on say entirety of some fictional universe bibliography, wh40k chatbot that knows all the lore!
"...it is very difficult not to extend trust in the same way as we would with a human."<p>a lot of research has shown that trust in each other, as humans, is quite misplaced because of bias, conflicting interests and other things.
There hasn’t been a decent search engine in the sense that one could craft targeted queries since Altavista. Google has been a chatbot since its inception. The competition is just showing that it’s no longer a very good one.
Counter-point: Yes they are. I’m already successfully using ChatGPT as a Google replacement for many types of searches.<p>You’re welcome to argue it isn’t a replacement. And yet my own two eyes observe indisputable evidence to the contrary.
ChatGPT still mines already existing information on the web. It kinda summarizes SERP results, kinda comparable to look trough first 10 results. I don't think it will compete with Google SE in anyway.
What I haven't seen here, is that AI chatbots remove the discovery process.<p>While you can get an option from a chatbot for your specific request, sometimes you can get something better while googling for possible solutions.
ChatGPT writes great first drafts, but needs a good and knowledgeable editor on the subject to create valid factual documents. Likewise, Google finds great pages with facts, but lots with untruths too.
Before ChatGPT's December 15th nerf update I was starting to use it in place of Google for technical questions. Something like ChatGPT will one day smoke what we consider search engines to be today.
Me:What is 123 times 456<p>ChatGPT: 123 times 456 is equal to 55288.<p>Me:That answer is wrong<p>ChatGPT:I apologize if my previous response was incorrect. 123 times 456 is actually equal to 56088. Please let me know if you have any other questions.
They're not a replacement but they will be generating so much credible indexable garbage that they will reduce the signal to noise ratio to the point that they are useless.
Mods need to edit the headline of this to match the title of the page. Nothing stating or implying this headline appear in the article and it feels like an attempt to troll HN.
No, they won't replace search engines, but they could weaponize the generation of SEO content spam to the point where search engines become overwhelmed and irrelevant.
I am substituting Google for ChatGPT whenever I have a question related to software development. There are many types of queries where ChatGPT is already a better choice.
This definitely is a threat to Google and websearch as we know it.<p>A lot of the discussion today is around the natural language output of ChatGPT. However, if that <i>presentation</i> is changed into a list of relevant facts, evidence, counterpoints etc., (based on which ChatGPT is generating Natural language textual response), we get an AI-powered search devoid of blog spam, SEO crap etc., Assuming advertising and commercial concerns don't corrupt the results with spurious results, this is quite encouraging.<p>Google is right to be worried about this.
But they have the potential. We'll see if it breaks into public.<p>People at some point were even saying Tikto will replace search engines.<p>Both of which can still happen.
In my view, this is an example of a problem with AI ethicists: they tend to naively overestimate the statu quo. In this particular case, the premise seems to be that people obtain reliable information from traditional search engines, hence chatbots cannot replace them because they are unreliable.<p>The truth: many uses of a search engine aren't even things where reliability matters much (what's the weather tomorrow? what's artist X's latest song?). For those where it does matter, results are unreliable as well, because the web is full of all kinds of crap and any attempt at sorting the wheat from the chaff failed many years ago. And sure, users can try to assess the reliability of the sources. Good luck with that (not you specifically, well-educated and tech-savvy HN reader, but at a population level). All those extremist propaganda sites that are climate denialists, antivaxxers, etc. have readers that think they are perfectly reliable.<p>Of course there are situations where checking the sources works (e.g. a scientist working on a literature review)... which, of course, are the situations where those people aren't going to trust ChatGPT anyway.<p>In summary, I think this kind of criticism widely misses the mark. The problem with this and other current issues is the corporate interests feeding us all kinds of crap, not whether the way of feeding it is more convenient or less.
well for me thay have already replaced, my first preferences. I used to google or go to stack overflow for resolving issues and woul need hours to come to conclusions. Now with ChatGPT i simpy start my research from there.
not sure if any openai people are reading this thread. but they need to figure out a way to add new content to the model. or have people submit their content. i got some answers that were not wrong but out of date.
why doesnt google just incorporate it into their search results? and then based on what people click, weight it more or less for each type of search? kind of like an ensemble search result
I tried baking cookies based on a chatGPT recipe… everything looked right glancing at the response so I dove in and started making them. About 2/3rds the way through it became obvious the steps were correct-looking gibberish.
A chat bot isn’t allowed to use external APIs or present things besides conversational text? Seems short sighted.<p>Todays chat bots should be considered dumb prototypes, when thinking about even the near future.
ChatGPT works very well as a tech demo but nothing like it will ever work at scale.<p>There simply is no business model for something as useful as ChatGPT, and there never will be. Let's assume for a minute that OpenAI will manage to improve their models such that the accuracy<p>There will probably be a business case for the chatbot Google will turn its search engine into, but it will be as far removed from ChatGPT as today's Google is from the good old Google of PageRank fame from 1999, which would show you exactly what you asked for, often enough to be considered genuinely useful.<p>The Google search engine from 1999 was a bit like ChatGPT is today: The demonstration of a new way of doing things that seemed, at the time, to be the way of the future.
Except we don't. PageRank is an awesome example of how computer science nerds can disrupt a market with their superior math but it didn't change the basic laws of capitalism and this is why it succeeded in turning Google into one of the letters of FAANG, but it failed at being sustainable. A Google-killing descendent of ChatGPT would have a hard time turning a profit without also turning into the same sort of thing Google is today: A convenient way of giving people ads instead of answers.<p>Here is my prediction:<p>1) ChatGPT and similar "AI" chatbots will never be more than tech demos<p>2) The Google of the future will have some shortish chatbot-copypasta for search results that resemble prompts to write essays<p>3) Google will sell ads for highly specialized and expensive services for the few people who need to copy-paste longer essays (which might or might not have a sufficient relation to reality) and are able to pay for it<p>4) The end-result will be virtually indistinguishable from the Google of today, because for every honest chatbot-aaS, there will be an uncountable amount of dishonest vendors of "free" auto-generated propaganda and spam, copy-pasted by school-children all over the world, because that will be the results Google will show below their own copypasta, which is just too short for this week's essay because the teacher knows about Google, too. And on HN, people will long for the good old times of 2022, when ChatGPT was, briefly, surprisingly awesome.<p>5) Your crush won't believe you that you wrote that touching love poem all on your own.
Let's go to the source ;-)<p><i>Question:</i> Will AI chatbots replace search engines?<p><i>ChatGPT Answer:</i> It is unlikely that AI chatbots will completely replace search engines. While chatbots are useful for answering specific questions and providing information on a wide range of topics, search engines are designed to help users find specific information that is widely available on the internet.<p>Search engines use algorithms to crawl and index the web, making it easy for users to find relevant information by using keywords and phrases. Chatbots, on the other hand, use natural language processing (NLP) techniques to understand and respond to user requests in a conversational manner.<p>In some cases, chatbots may be able to provide users with the information they need without requiring them to use a search engine. For example, a chatbot could provide users with personalized recommendations or help them find specific products or services based on their preferences. However, search engines will likely continue to play a crucial role in helping users find information on the internet.<p><i>Question:</i> Imagine an scenario where they replace them almost completely, what innovation would be there?<p><i>Answer:</i> If AI chatbots were to replace search engines almost completely, it is likely that the chatbots would need to be significantly more advanced than they are currently. In order to replace search engines, chatbots would need to be able to effectively index and organize the vast amount of information available on the internet, and be able to provide users with accurate and relevant results in response to their queries.<p>One potential innovation that could enable chatbots to replace search engines is the development of more advanced natural language processing (NLP) techniques. Chatbots would need to be able to understand and interpret complex queries and requests, and be able to accurately provide relevant information in response.<p>Another potential innovation is the development of more sophisticated machine learning algorithms that can help chatbots understand the context and intent behind user queries, and provide more relevant and accurate responses. This could involve the use of techniques such as deep learning and neural networks, which are designed to enable machines to learn and adapt over time.<p>Overall, it is likely that the replacement of search engines by chatbots would require significant technological advancements in the field of artificial intelligence and natural language processing.<p><i>Question:</i> Do you think technologies such as ChatGPT will confuse search engines to know if web pages are written by humans or computers?<p><i>Answer:</i> It is possible that the development of technologies such as ChatGPT (short for Chat Generative Pre-training Transformer) could make it more difficult for search engines to determine whether web pages are written by humans or computers. ChatGPT is a large language model that has been trained to generate human-like text by analyzing a large dataset of online conversations. As a result, it is capable of producing text that is difficult to distinguish from text written by humans.<p>However, it is worth noting that search engines have become increasingly sophisticated over the years, and they have a number of techniques at their disposal for determining the quality and relevance of web pages. These techniques include analyzing the structure and content of web pages, evaluating the credibility and authority of the sources that are cited on the pages, and looking for patterns of suspicious or spammy behavior.<p>It is therefore likely that search engines will continue to be able to identify and prioritize web pages written by humans over those written by computers, even as the capabilities of technologies like ChatGPT improve.
they totally are, not 100 percent but mostlz working already. risks of misinfomation will further plummet, and in already living with it for eg programming questions<p>this bot even references sources, that's where it should go: <a href="https://you.com/search?q=what+was+the+recent+breakthrough+in+fusion+research%3F&fromSearchBar=true&tbm=youchat" rel="nofollow">https://you.com/search?q=what+was+the+recent+breakthrough+in...</a><p>ask "source pls" if it doesnt give you one
Assuming this refers to <a href="https://chat.openai.com/" rel="nofollow">https://chat.openai.com/</a>, it's rather difficult to try it out. After creating an account it demanded a phone number. Of course, I will not supply my own, and temporary ones I tried were either recognised as such or rejected has having already been used.