YouTube Video
Episode Summary
– Articles, videos and podcasts all over are claiming ChatGPT spells the end to Google Search
– I share some perspectives on why I do not believe this is true
– Google has their own large language model named LaMDA (that was famously said to be conscious)
– Google indexes the world’s information in real time and has direct access to the information required to train these models
– They also have some of the world’s greatest technical talent
– Google also has billions if not trillions of searches and can use those searches and signals from success versus failures to train their system
– Yes, there is possibly some impact on the validity of Google’s current advertising model if people seek individual responses instead of the current search engine results page
– However, people are accustomed to free and OpenAi will have to figure out a model too
– OpenAI has deep flaws including data only to 2021 and the inability to include links as one example
– Currently, no strong formatting or image inclusion to improve understanding
– ChatGPT is over capacity and not functioning several times since launch already
– Look at the speed Google released Imagen and Phenaki to show what they want to do when motivated and inspired
Resources:
Will OpenAI’s ChatGPT Kill Google? – YouTube
Jason’s big night, the culture war in tech + Google Search vs ChatGPT and more | E1635 – YouTube
ChatGPT END-GAME: OpenAI vs. Google vs. Amazon vs. Meta – YouTube
ChatGPT versus Google and the future of search
Google: ChatGPT Is An Overstated Threat (NASDAQ:GOOG) | Seeking Alpha
Can ChatGPT really challenge Google? Well, it’s complicated.
ChatGPT smoked Google. Here’s why the search giant didn’t release an advanced chatbot first.
OpenAI’s ChatGPT Vs Google’s LaMDA
Is ChatGPT the next big threat to Google’s dominance in the AI market?
ChatGPT Should Worry Google and Alphabet. Why Search When You Can Ask AI? – Bloomberg
OpenAI’s ChatGPT chatbox gives competition to Google Search | Fortune
(1) chatgpt vs google search – Twitter Search / Twitter
Overview – OpenAI API
ChatGPT
LaMDA: our breakthrough conversation technology
Google engineer Blake Lemoine thinks its LaMDA AI has come to life – The Washington Post
Imagen: Text-to-Image Diffusion Models
Phenaki – Google Research
Hashtags
chatgpt,chatgpt vs google search,openai,google,alphabet,openai vs google,lamda,imagen,phenaki,large language models,chat gpt,chatgpt explained,artificial intelligence,openai chatbot,chatgpt coding,machine learning,chatgpt examples
Automated Transcript
Chat GPT versus Google search. Chapatti came out, took the world by storm, and everyone loves it. Playing around with it, asking questions, making Seinfeld scripts, writing code, all of these great things. And then the articles emerge. Google got smoked by Jack GPT.
Is Chad, you butt the next threat to Google’s dominance? Google faces a serious threat from Jack Chat, GPT and in this video. I just want to sort of question the, you know, why is this the sort of sort of perspective that is happening there? There are some great sort of articles and pieces that maybe it’s not such a clear threat to Google’s. Dominance but I just overall it was really interesting you know sort of this soon as people sort of interacting with chat GPT all of a sudden saying Google search is dead and you know for many reasons I think this is definitely I err on the side of this is an overstated threat and I’d like to share why first of all just take you know continue to sort of take a journey down that path which is.
You know, people saying, hey, I did a question on Google search and then I typed it in on chat, GPT and ohh man. The other idea is, hey wow Chad, GPT gives you such great answers and it only gives you one answer and that’s fantastic, but. You know, what I’m struggling here with is just sort of the bigger picture of this and I’ll I’ll get off my pedestal and just talk about. A couple points. So I mean first of all open AI funded, you know, a billion dollar investment from Microsoft, incredible team. I love you know, the innovations that they’ve made and also some of the, you know, focus that they’ve brought on these systems in the world today. However.
They are interacting with information the way that they are training their system is through the world’s information and who organizes the world’s information. Yes, Microsoft has Bing, but generally most people when thinking about that, think about Google and Alphabet. We know, I know that you understand that mix and so they’re like using, you know, I’m guessing some sort of scraping mechanism from that’s even leveraging information off Google search itself to train the models that then create chat. GPT. So it in itself is reliant on Google search. I’d like to also sort of talk about just the technical talent. Obviously the open AI has fantastic. They’ve been able to execute this on a long time. But if I go to, you know, chat GPT at this exact moment, it’s down, it’s that capacity they’re struggling to handle the the weight of usage of the system where Google has been running relatively smoothly.
For years on end, not just Google itself, but to YouTube, Google Maps, huge multimedia application serving billions of people every single day with barely any downtime or leg or problem. And yes, maybe you’ve experienced some yourself, but overall, I think we’ve got to admit it’s pretty good. And from a downtime perspective, what we’ve seen from chat GPT versus Google. And again, I know they’re early stage companies. We’ve just got to think of the scale that Google has been.
Operating now, one of the other super interesting pieces here is that Google actually, you know, it grabs it is grabbing information in real time and organizing that information. And if you’ve experienced chat, GPT, you can see the fundamental limitations. And I guess the idea here is that they have only trained their system until 2021. I’m not sure when in 2021. But as an example, I did a query around Adam Newman of we work and it gave me this wonderful.
Description of how Adam Newman was a success story and how successful we work was and my understanding, I my guess is that it didn’t have the information on the downfall of we work that had actually happened after the model had been trained. And so it just gave me the information that it knew at the time. Whereas Google has real world, you know, access to this information and can train that model that they’re generating in real time. And what is that model? We’ve you know, we’ve seen this talked about publicly. It’s not like people think like hey, it says chat you. I don’t know why it’s taking such a negative tone here. I just think this is really short sighted. There’s some really smart people who are talking about this. I think they really, I mean I don’t think the really smart people are saying hey Google is done, but I think there is this sort of hysteria around it that I just think it’s really sort of short sighted, not understand, I don’t know, just not comprehending the scale that Google is operating, the fact that they already have.
Their own version of this. Yes. It’s not maybe as publicly accessible as the way that open I release it. But like we’ve already had people internally in Google interact with this system and say that it’s conscious. So it’s not like Lambda. Google’s version of this is you know dumb and and you know I just don’t think, you know we’re thinking about how easily that you know, Google can execute on this and I think this also speaks to something that we’ve seen.
You know over and over again is where when you have like a scale that is as big as Google, how quickly you can deploy technologies and have mass adoption at speed. And the one example that I think about is Google Docs did a summarization feature where it would auto summarize what was taking place in the Google Doc, which was, you know, which was a company thing that companies were building entire business models off. They released that immediately deployed that into Google Docs. And then you know reaches touches billions and billions of people immediately and so that’s the same functionality that they have here. There is nothing stopping them from embedding. I mean there may be something stopping them. I’m guessing it’s related to the business model in Google search and I think you know I think that’s where maybe there are some conflict here which is.
Google is incentivized to, you know, to serve up multiple results on the search engine results page and then multiple of those are often, you know, paper click advertising and that is a basically the prime driver of Google’s business and that is a massive business. And so they are incentivized for that not to go away. Um. Now, so to me that’s the biggest challenge. It’s not the technical challenge, it’s not the access to information, it’s not the talent that they need to make this happen. It is fundamentally, how do they incorporate this kind of say language interaction or information interaction without cannibalizing Google’s business model or the vast amount of customers slash partners, however they look at them during that process. So I’m guessing that this is one of.
Sort of Google’s core. I mean that sort of one of their missions, maybe not core that they’re working on figuring out especially as they see the adoption the excitement of chat GPT here and. The other thing that I wanted to then sort of point out to just sort of I think support this case is the. Like just at the speed that they can, they they can move at. So for example within, you know, a day or two of Facebook.
Releasing their meta sort of text to video. This is actually Imogen images but Google released imaging video and then they said hey let’s go up a knot up a notch and we’ll do Funaki. I still not sure if that’s how it’s pronounced but let’s synthesize video multi prompt so that they can create continuous video over time with understanding that can tell a story. So like I I think what you know what we’ve seen here is. Guys, stable infusion rows is open AI rose. First of all, Google seeing people searching for this, exactly what they’re searching, what they want to know, what they want to accomplish with it so they can execute on that. They have the technology to have the capability to have the servers, all of that. And when they are motivated and inspired, they can move at incredible and speed to deploy those technologies and reach people at scale. And there is opportunities with this obviously with open AI embedded into Microsoft and Microsoft Teams and all of this stuff. But you know, Google has that.
Strength. And you know, even though Microsoft has invested in open AI, there’s maybe still a little bit. Alignment that’s required there to then release it directly into systems. There’s a couple other things that I’m just thinking of which is. We, you know, we.
Are liking the responses that chat GPT is created. But in general, like even as I’m looking here, it’s like this article is telling me information which is meaningful, but it’s also giving me video and image context to help me understand what I need to understand better. GPT in its current state is lacking that part. Of course there’s this sort of idea of pairing it with Dolly or there’s other ways that chat GPT can be enhanced to include meta information and and all this, but then does it just cannibalize it and. Become another version of Google search. Like these are things that were not necessarily thinking about whereas even in a how to guide it’s great to get listed out description by Chachi PT. But I want to see some images along the way of how to pour properly or whatever it is. Whatever shitty example that I’m thinking of right now and just in a general piece here like chat she GPT and I have somewhere in the quotes but just Sam Altman talking about the compute cost of chat.
CPT just being ridiculous, right? Yeah exactly. We will have to monetize it somehow at some point because the compute costs are eye watering as and yeah great million years, a million users in five days that’s, that’s fantastic. There’s lots of good signals that this is a great time for open AI that they’re doing some incredible work. But I just, I just wanted to sort of tackle some of these I think hyperbolic overblown claims that this is an end to Google search.
They will adapt, they will continue their still one of the best businesses probably ever built in the world. They’re not stupid people and they have a lot of resources and a lot of talent. But to execute on this and they’re already ahead seeing signals in real time of where they need to guide this from a product development, from a marketing standpoint. And I don’t think we should discount them on this journey. So I love both. I love open a I want to see them continuing moving forward. I want to see Google and Alphabet continuing.
Moving forward and you know building this incredible applications and functionality using machine learning and large language models to do just mind blowing stuff and I think that will continue in parallel. There will be at least in this current time no destroyer of each other and if there was one to be destroyed here, I think open AI would be the one who is I guess David fighting Goliath even with the massive amount of resources. That they have. So let me know what you think about this. Would love to hear. Maybe I’m stupid, maybe I’m wrong. That would be great to know why I always loved learn. I love the feedback loop as always. I’ve got some great resources here and I’m going to continue to talk about content like this. There’s a lot more stuff happening right now that’s super interesting. I look forward to creating some videos on so. I hope you have a wonderful rest of your day. Bye bye.
