The funny thing is that Google was dragged for showing a wrong answer during their demo, but Bing did too. They just didn't get caught. The "things to do in X" recommendations didn't make any sense and / or were wrong with their descriptions, the pro's and con's comparison showed points that had made up information, and their summation of the financial report was flat out wrong. Everyone just jumped on the hype train before Bing was called out. I think these will be incredible in the near future, but there's going to be a lot of issues raised around the accuracy of the output. It's a bit scary when you see how convincing it their output sounds despite being absolutely incorrect.

5:05 Reality of actually clicking links 9:15 Twitter issue 9:55 recursive problem 11:45 Web 3 13:08 Verge 15:45 Google error

Something I think this discussion missed is that in within a year or so, most “SEO optimized” articles will be entirely written be generative AI in the first place. Both their cost and value decrease in lock-step

the new BIng is incredibly interactive, you can do research through it by asking very specific questions, like "make a table of the best oled tvs ranked from lowest to highest price, include resolution and brightness as columns"

I think the bing+chatGPT combination may result the replacement of the blog sites with more vlog sites such as KZfaq. Another thing that can be assumed is, if Microsoft could persuade the bloggers or blogging agencies to write verified articles for undeniable incentives exclusively for them which will be used to optimize the chatGPT subsequently, Microsoft will be able to pull the strings wisely while promoting the advertisements related to the content it generates.

The argument at 9:40 is exactly what I was hoping someone would bring up. An AI search engine needs the new content to feed off of or it becomes irrelevant. If the AI search engine is not supporting the content it consumes, then the AI dies.

5:15 Bing chat isn't ChatGPT, it's more powerful and behaves differently. The rumor before the Bing chat even came out was that it would be based on GPT-4, while ChatGPT is based on GPT-3.5. The official word from Microsoft is that it's based on a language model that is "even more powerful than GPT-3". In an interview from Q4 2022, a high-level OpenAI employee said "Hopefully in 2023 we'll forget Dall-E 2 and GPT-3 and focus on something new", but that was for all of 2023 and we don't have any exact date for GPT-4 yet.

"We'll call it ChatGPT" (5:10): It's not Bing search or ChatGPT but a hybrid called Bing Chat. I know you guys realize that, but let's not gloss over a couple years or more of hard work by Microsoft engineers who contributed Bing integration, Azure infrastructure buildout, and most importantly, massive amounts of additional LDM training and filters to help transform raw ChatGPT into something closer to broad usability. Microsoft didn't just write a huge check to Open AI and sit back waiting for it to be finished. Obviously lots more work to be done, but it is still in pre-release after all.

I have so many questions :) Old model: The user executes a search via a search engine that lists pages, and the user has to visit the pages; this generates traffic that translates to advertisement revenue. Revenue goes partially to content creators. Now: The user executes a search via an IA search bot that processes the content itself so the user will get results without visiting the platforms/web pages. It is more convenient for the user, so most users will roll with the "pre-processed" results they get. It's a downward spiral where less revenue for content creators causes a decline in original content available, making the search less effective. Content creators may hide their content behind a paywall (premium plans, freemium model, etc.). Can a big company pay for each premium plan to make their search more relevant? Is it possible to separate human search from AI search? For an AI, you don't need to load any "fancy" part of your site, just the actual text, so it's cheaper. The IA visits are more frequent and way faster/shorter than human visits. It makes sense to me. Can content creators use AI to create content? If so, how it won't degrade the quality of the search results? Can it create an echo chamber for AI with more and more homogeneous content over time? How to prevent this? If the search AI's primary goal is to give me relevant content, our interaction trains the AI to determine what content I find "relevant" (or, more likely, pleasing). Will that create targeted profiling? How would I know if the search results are based on facts a reliable statistics or if it's just favorable for me? Mom, pick me up! I'm confused, I'm scared, I'm trapped :)

I think it's a real issue. One potential solution to this dilemma is for websites to focus on providing additional value beyond just information. They could focus on building communities, offering personalized recommendations, creating interactive experiences, or providing unique insights and perspectives that cannot be replicated by an AI. This way, they can continue to attract users and generate revenue even in the face of AI-powered search engines. Another solution is for search engines and websites to work together more closely. Search engines can partner with websites to share revenue generated from search traffic or offer other incentives to encourage websites to continue providing valuable information. This would create a more symbiotic relationship between search engines, language models, and websites, ensuring that all parties benefit from the growing use of AI-powered search. 🤷🏽‍♂️

Recipes will be huge, because I’m tired of scrolling past the 2,000 word essay about how the chefs grandma studied at Yale and traveled to Spain but one day she fell down a well on a cold winter in 1894 and forgot how to cook but relearned and made this one recipe better than before.

This is likely to break SEO and the whole concept of free websites, and I think that's overall a good thing. We've gotten used to the advertiser driven attention economy as though it's good and natural, but there's plenty wrong with it and it's about time it got disrupted.

Maybe Microsoft, ChatGPT etc will start paying content creators to keep the internet updated for ChatGPT to use. If ChatGPT uses info from your content, you get paid.

The title of the article had nothing to do with what they discussed, which is another topic entirely as we discuss the integrity of AI but I also feel like this particular topic was covered too early and lacked the facts of how the new Bing actually works. It's clear that no one on the panel has actually used the new Bing (I just got access a few days ago as of this writing). They kept talking about requesting a list of the top 10 TV's and suggested that it's just plain text without links when the reality is that every answer from Bing's chat bot is a link. Almost none of it is just plain text but rather text sourced from a site. And based on my use it's almost impossible to just highlight the text because it's all tied to a variety of websites. While that doesn't dismiss the theory that people will just accept the information provided without drilling down, it does disprove their take that the Bing Chat feature doesn't provide linked context with its answers. Secondly, it's also clear that no one on the panel actually uses Bing or the Edge browser. I've been using Bing almost exclusively for the past 4 or so years and have been using the Edge browser for work for nearly 2 years. I'm not here to shill for Edge but in my time with it I've found it to be better than Chrome but when it comes to search one of the first things I noticed was how Edge and Bing provide search results compared to Google. When you click on a link in Google Search you're taken to the page you clicked on. Sounds right but what happens if you want to go back to your search results? You'll have to click/swipe back, which could be multiple clicks/swipes if you've travelled through the website you clicked on. With Edge, however, the link you clicked opens a new tab while your search results are retained on their own tab. Moreover, if you click/swipe back the tab you were on closes and it goes back to the tab with your search results. Having my search results retained wasn't something I thought I needed until I started using Bing and now I can't go back. I still use Google from time to time (mostly Google Maps) but Bing is my primary search engine 99.9% of the time My point is that Marques mentioned that people will get their answer then just leave but that behavior is a little different with Bing and Edge if/when you decide to click on the results. Anyway, I'm sure they'll revisit this topic once one of them (or all of them) have had time to actually use the new Bing and a lot of what they discussed here will be revised and hopefully corrected based on their initial thoughts.

Publishers would move from ad revenue to Information revenue, i.e. every time chatGPT uses publishers work to come up with an answer it will be required to pay a small fee to publisher.

It's too early to tell if Bing is winning the AI war, but they've definitely fired an opening shot across Google's bow!

The problem of content creation vs. content needed for AI to reference is really the main issue we don’t have an answer for yet. Exciting times. The „like“ count is too damn high ☝️

I already stopped using Google since the new Bing (after signing into the waiting list it took me a week to get access) is cool and informative with this useful Chat AI function which even writes in my native language (which has a small users base), so at least for now Bing wins - FLAWLESS VICTORY 😀

"Read a book, Andrew" is one of my favorite parts, loving all the discussion around AI bing/google search

Because of the controversy regarding its advice, which is why we want AI, it’ll work best when individual groups, businesses, and persons are able to have access to or own their own AI that they are able to download content into. So much copyright infringement‘s too. It was a big problem at the gecko, copyright infringement’s. I see AI purchase packages down the road, that have packages of data pre downloaded into it, and options for people to create/add content/data into the AI bot.

I just got early access to the new Bing and its INCREDIBLE. Much better and more comprehensive than ChatGPT. The new sidebar in Edge is a game-changer too.

I feel like Marques brought up my thoughts on this exactly. A good portion of consumers won't just take the AI answer at face value most of the time. Not that AI can't decipher some of the depth too but there is so many more questions to be asked. Will an AI know how to tell the difference between the article with top 10 tvs posted in 2023 vs 2020? Even deciphering if a tv from 2021 is a better option than a 2023 tv. Also how many people will search top 10 4k tvs to buy, or tvs under a certain price point, and so many other variations of "top 10 tvs to buy". I just see there is a lot of very variable information that's just being given value based on the ai's base knowledge without any actual system to grade. This is a starting point but I think it'll need to get to the point of genuinely look for a tv's specs and through ratings of tvs. For example going to RTings, Consumer Report, amazon, Best Buy, Walmart, etc and making analysis based on the information from professional and consumer reviews and ratings. Even valuing a consumer review in correlation to a professional review. Then knowing how to value that differently or the same based on unique situations. For a trip itinerary eventually letting you choose variables like pricing range, deals, and check thoroughly through paid and customer reviews, and choosing or allowing me to choose different aspects of the trip. Then for recipes not just giving me an Apple crisp recipe but letting me choose if I want something completely made from scratch or using a premade crust or just using Apple filling vs fresh apples or what style of toppings. Even to the point of letting me choose like consistency and taste aspects of the crisp. So again a it's a starting point but there' s so much more depth it'll need to be something reliable day to day.

We need citation links (not unlike a research paper) back to the source of the information that generated the Chat GBT response.

Couldn't care less for seo, ads, affiliation, etc. It is not a consumer but business oriented stuff and whoever works in those fields will have to adapt. I imagine bing will have to make search profitable in some way, but there has to be a strict line between a useful link and an ad, which are thrown in your face on google

I switched to bing yesterday because of the microsoft rewards fidelity program and I asked how long horses live and it said 25 to 30 years according to 5 sites, and it showed the parts of each sites that showed 25 to 30 years so I didn't even bother looking at other sources and entering the sites.

Marques has been in the tech realm most of his life, but he’s just discovering A.I. I’ve been following it since Ray Kurzweil’s 2012 book “How To Create A Mind.” Then when Deepmind beat Go, all bets were off. GPT-1, 2 then 3 was actually mind blowing.

Chromecast with Google TV is the best accessory to turn anything into a Google TV. Google is straight up the best company.

In a year or so we just ask the AI to solve this. It's essentially not an advertising issue. It's having money to pay rent, buy food and have free time etc. Better options like universal basic income, free energy, and food synthesizing. All systems are about to change. Awesome time to be alive right now!

Whoa, that intro had me hooked right from the start! The first few seconds had me feeling like I was about to set sail on a high-seas adventure, and then it hit me - it sounds just like 'The Bounty Main Title' intro! It's amazing how a few notes can evoke such a powerful sense of nostalgia and excitement.

Hopefully this will mean companies will go from writing awful articles to their site aesthetics, to writing ai readable databases for search ai to reference and demand a cut of the ai's click-through returns.

This video was more of an argument for promoters and the economic impact AI will have on them because I might not view their content... If only I had something that could scan the video first instead of me wasting 20 minutes. They went over which sites had the worst answers out of Google, Bing or ChatGPT they they searched searched the best TVs, the best banana bread recipe, and which site gave the worst wrong answer... The irony forced me to leave a comment. I am still a fan! But, In my opinion, it scares me to think of other countries having this technology and America banning AI. There is a lot of doom and gloom at first. It is shocking the more you think about it. When the shock wears off go back to the drawing board and find ways to innovate or even create a better business model. Good luck my friends.

First step that will happen if a chat bot steals traffic is that sites will block the crawler leading to bad results that way.

Regarding loss of clicks to free site creators, couldn't Microsoft build auto-clicks into their Chatbot's searching capabilities? So, it basically gives clicks to all links being used in its answers?

This conversation is waaaay more educational than just learning about chatgpt. Wow !

Finally someone is talking about this. This is a giant problem and it’s getting glossed over.

The new Bing browser is too chaotic. I quit after a few hours and went back to Chrome. It needs to be simpler.

10:11 that is a very good point, David.

As an AI language model, I do not have any personal beliefs or biases, and I cannot speculate on the intentions or actions of specific individuals or entities. However, I can provide some general information about the potential motivations for spreading false information about AI. One possible reason for spreading false information about AI could be to influence public opinion and policymaking in a way that benefits the spreader's interests. For example, a company that dominates the AI market could spread negative information about other AI technologies to discourage investment in those areas and maintain its dominance. Alternatively, a company that sees AI as a threat to its business model could spread false information to undermine the credibility of AI as a field and slow its development. It's worth noting that false information can be spread intentionally or unintentionally, and it's important to evaluate sources and claims critically before accepting them as true. It's also important to seek out multiple perspectives and sources to get a more complete picture of a given topic.

this video came QUICK after they mentioned you as the best KZfaqr

Every manufacturer should think. How is this better? If they can't make anything better. They can always make waterproof shoes, shirt, socks, bag, laptop, phone, VR water proof headset that charges with light, Waterproof headphones, waterproof shorts where you can put stuff inside (zip ties stiching). Instead of building redundant stuff.

I previously thought I was the only one worried that web advertising is screwed by AI, but now I feel better since many people asking the same question if we can afford free web after the lunch of Bing chat.

In my opinion, if I want to know what the best product is I will always look into link to see why it is the best (but maybe not for me)

I mean, I barely go to websites anymore. Every time I want to know something or search for something I come to KZfaq instead. That's a win for Google, although kind of ironic. I'm pretty sure that KZfaq massively decreased traffic to many websites

Chatbot doesn't give you the questions to ask though. I don't look at your stuff to decide what to buy, I do it because I learn things I didn't think to ask.

Bing is definitely the underdog, and the underdog tends to do better because they’re more hungry and determined to get to the top. Where as with google is already at the top, so they’re comfortable and now probably won’t do as good. So I think Bing is definitely getting better

It’s exactly the same story as with google news. Publisher were complaining that google copies article and takes whole profit for himself. That’s why a linking-tax has been proposed. It would works as well with AI

just switched to bing and the new edge in past week, im actually surprised how decent it is, lmao😅

Wonder if advertiser's will pay to be suggested by AI soon

As someone who buys top 10 TVs I am guilty of chosing the top 1 and not looking further. I do this because I don't need the best TV of all time, but if I pick the best one in a list it's probably going to be decent. If I really wanted a good TV I would be looking for a specific product and would not use AI search.

Used both. Bing/Chat GPT gives a sharp and smart response. Google is much more glib, loquacious and seems wordy. Also, Bing is winning with great user interface.

I think the next big “SEO push” will be just outliving the competition. You’re totally right - where’s the incentive to keep writing quality content? Well, if you’re the ONLY one writing quality content, and the primary source for Bing/Bard…that seems like incentive to me.

bloggers not creating new content due to no incentives, well one pragmatic solution would be ai generating transcripts of youtube videos, which they are very well capable of, but i wonder if such computing power exists which computes all this massive history of youtube transcripts in seconds, perhaps in future.

Can't wait to try it out by myself

At one time Google was going to scan every book ever written in the history of humanity and make it digitally available. What a wonderful thing, but then there's the authors and publishers who need to be compensated for creating them. So that was largely the end of that. Our economic system needs to catch up with our technology.

On asking ChatGPT if it will not change the world if ChatGPT is combined with a search engine the response was: Combining the power of an AI language model like ChatGPT with an internet search engine could potentially lead to improvements in the accuracy and relevance of search results. AI language models like ChatGPT can process large amounts of text data and generate natural language responses, which could be used to enhance the quality and depth of search results. However, it is important to note that there are also potential ethical concerns and challenges that come with using AI in search engines. For example, there are concerns about bias and discrimination in the algorithms used by search engines, as well as concerns about data privacy and security. Additionally, while AI language models like ChatGPT are powerful, they are not infallible and can make mistakes or provide incomplete or inaccurate information. Therefore, it is important to use a combination of AI and human review and oversight to ensure that search results are accurate, relevant, and ethical. what an answer!

AWESOME 👏🏻 New sub here. Cheers from Western Australia 🇦🇺👍🏻

I’ve used bing and I hate how they lobotomized it. The first two days was wow. Now I can’t even use it. 😢 I hope I get to use the future.

Thats what Chat GPT says... As an AI language model, Chat GPT can provide you with information and insights about electronic gadgets based on a vast corpus of data and knowledge. Chat GPT has access to a wide range of sources, including expert reviews, user feedback, technical specifications, and industry trends. However, Chat GPT is not perfect and may not always have the most up-to-date information or be able to provide specific advice for your individual needs. On the other hand, KZfaq reviewers can provide you with first-hand experience with the gadgets, and you can see how they perform in real-life situations. They can also give you more detailed information and demonstrations about specific features and functions of the gadgets. Ultimately, the decision of whether to trust Chat GPT or a KZfaq reviewer for purchasing an electronic gadget will depend on your personal preferences and the specific information you are looking for. You may want to consider both sources and weigh the information they provide to make an informed decision.

In the gaming community, there are people who dedicate their time and effort to creating mods and guides for free. If publishers stop making clickbait titles, then only those who genuinely want to help others will continue making articles and lists with recommendations.

Actually Bard was correct but the grammar was misleading @15:44 The telescope did took the first picture of "a" specific planet, but not the first of all planets

Here is a thought... SEO now becomes AIEO in the future in order for the AI to include their content as the model becomes: if the AI uses my content for its model/source I get a kickback as content creator... congrats everyone, we will soon be working for the AI.

yes, yes and yes. its that simple, better adapt and start thinking bigger

I think for the better AD business and Content creator/consumer ecosystem, they will find a proper business model (or eventually will) to balance this. Or maybe a brand new business model will revealed

You can say, the AI visiting these publishing websites is a win regardless of it not being human. As long as there's traffic. In fact, this could be a win win

Firstly, having used chat gpt, one of its strengths is it is conversational. So, I believe browsing through links after you get an answer may be a last resort, not the second thing you do. I found it much easier to just ask a follow up like "Can you update the plan without any museums?" Secondly, we are not sure how bing plans to monetize this. But one way to get out of the vicious circle of publishers not getting traffic and stopping publishing is Microsoft potentially revenue sharing with publishers. That will disrupt Google even more as they don't do that with things like News and the summary on top.

Google has the real power in this. They can literally take the auto captions transcription generated for each video on KZfaq, and literally train their ai to search across everything including videos, text, and the web. That's going to be the most powerful form of AI.

The problem is that if you reduce the number of clicks, by any amount, it means a significant amount of advertising dollars are not going to that site. If 10% of people stop watching videos or clicking on sites, then 10% of your staff are getting fired. If even 10% of people are lazy and don't follow up looking into more info, then you lose that revenue. David's point is pretty important though. It's almost like there needs to be a tax on GPTs to fund the creation of the content that they are ingesting.

I feel like the best solution from a commercial point of view for both Bing/Google and the pages they get information from, is to paywall the AI. The majority of people will still go to pages, but a lot of people buy into the AI generating revenue for Bing/Google. From a consumer point of view this could cause quite a difference in service between people who can afford ChatGPT/Bard and those who can not.

FYI - Your link to the full episode is not working on my end. I know how to get to it but that link is broken.

I think there is no going back now, even if if the delivery form changes somewhat. The internet dynamics will probably significantly change now to adapt

I honestly think this will push many outlets to move to the web 3 model and charge the ai to query. Ai will generate ad revinue and a digital transaction will happen.

An epiphany just popped in my mind. This just gave me some fresh inspo for my next upload. Can you imagine like, 2-3 months from now, what pivot will we be really in?

If art generator models have taught us anything is that we should avoid using this phrasing of "in the style of "

You are asking the wrong questions. Instead, we should focus on discussing how to update the current model. For example we could consider compensating website creators on a monthly basis for their contributions to AI results. It is imperative that we shift our focus towards the future and not dwell on the past. The prevalence of AI is here to stay, and we must therefore adapt the current regulations to allow for data creators to receive a fair share of the multi-billion revenues generated by big tech. A significant portion of these profits should rightfully belong to those who create the data.

I think the best way to get around this for these review sites is to change the way you grade the products. Like a color coded and multigrade scale, just complicated enough for them not to scape it and can replicate the results.

Isn’t chathpt being hyped like cortana? How many people are using cortana to ask it “do I have a meeting today?” Can cortana give you the right answer? Yes but why aren’t people using it? If they crack this Chatgpt will be useful otherwise I see people being over optimistic Chatgpt

Love the mic arm stands. Will like to know the name. Thanks

People forgot that Google also disrupt the traditional ad business globally.

It won't be too long before AI chat models transcribe and ingest audio and video, and incorporate it into their text responses.Or are they already doing it? Is this a concern for mkbhd and his coworkers? Even now, I would understand if you and others stopped including the transcriptions of your videos on KZfaq.

I think the link to the full episode is broken. It redirects to something called askmarket i think.

The really big challenge here is having people try it out. Once they do a majority will get hooked. That's how a killer-app works.

I think the world is headed to simplicity and efficiency when it comes to search engines. I know many publishers and websites will lose lots of money but I think the way for them to make it fair that when bing Ai or google Ai in the future should compensate publishers every time they use their data to answer questions. Sorry for my English.

this is gonna cause a content-pocalypse, if websites aren't able to get to the top. we started with people posting stuff to the internet because that was the future. then we had bigger services, requiring bigger infrastructure, and donations couldn't keep everything running. from that point on, the internet became fragile.

I think one potential solution would be for websites to list Ads structured data (JSON Schema) for each of their pages that Bing/ChatGPT would have to display when an answer it spits out is mostly attributed to a couple of websites 🤔 That said, it would make the experience of chatting with Bing/ChatGPT quite spammy in a way.

This is eye opening. I'm always careful about information on the internet but will be even more diligent now. I love these waveform clips.

Google Assistant is getting worse. This is what I've had over the last couple months. And asking these in a different way did not resolve the issues: I ask if it will rain: "Although there is only a small chance of rain on Tuesday, there is a very moderate chance of rain on Tuesday" I ask the temperature: "Today it is in the mid 70 "S" 70 S? It is 70's I ask a question about carbonation: " carbonation is COlorado 2 dissolved into a liquid" Colorado 2? It's CO2. I ask to convert 1 inch to millimeters: "There are 39370078.7 inches in a million meters.

They should have used Clippy as the ChatGPT interface.

2:55 AI also filters out "bot articles" or those annoying regurgitated "top 10, top 5" listicles.

I think that more sites are going to require a login to access the body of their content to prevent them from being cut out of the loop.

I do SEO,and I strongly disagree with some of the comments. On the questions in minute 8, search results without clics, there are some statistics that talk about this. Also its a threat to google since you can ask ask chat GPT to read, catalog and resume a video transcription. Anyway love the show, huge fan here

It's once again the question of how any industry status quo is so ubiquitous that people who are locked in would argue against technological advancement that challenges it - do we advance technology no matter what or do we preserve the industry no matter what; and what is the middle ground...

10:15 and THAT was my query about this “new” tech

Google is actually making the smarter move by letting their competitors make mistakes first so they can learn from them and outdo them.

If publishers stop publishing information because of low interaction, search engine companies would be forced to make their AI more general purpose. Now you get an AI that would instead of taking context from webpages, would "watch" movies ask me some questions then recommend movies. And if directors stop making movies, you get Seinfeld.

im not saying you are wrong but there will definitely a shift to creating content. companies, products, vacation hotels, etc will still do content and people will still review it, but review sites, news, aggregators, and such content will be impacted the most

I can’t wait to see what happens when these text models are combined with smart voice assistants and AR/VR technology 🤯🤯that for me, is what Web3 should be about

When will be able to use this without entering a wait list?, or configuring Windows?.

I would be very fine with Google copying this feature verbatim, since Bing has been copying the ever-loving-heck out of Google Search down to the millimeter for so many years.

Yes. Please extract information from websites that answer my questions in the search. I HATE clicking endlessly on links that have nothing to do with my questions... Seriously! Thank goodness for AI sifting.

Props to you guys. A conversation like this needed to be had. People are panicking about being replace by AI right now but AI is just an efficient tool who needs expert humans to help other humans. So there you have it we are not replaceable just yet. Maybe 20 years from now.