Discover how Artificial Intelligence is revolutionizing search engines, shaping our access to information and paving the way for a more knowledgeable society.
Guest: Eric Olson, Co-Founder & CEO at Consensus.app [@ConsensusNLP]
On LinkedIn | https://www.linkedin.com/in/eric-olson-1822a7a6/
On Twitter | https://twitter.com/IPlayedD1
_____________________________
Host: Marco Ciappelli, Co-Founder at ITSPmagazine [@ITSPmagazine] and Host of Redefining Society Podcast
On ITSPmagazine | https://www.itspmagazine.com/itspmagazine-podcast-radio-hosts/marco-ciappelli
_____________________________
This Episode’s Sponsors
BlackCloak 👉 https://itspm.ag/itspbcweb
Bugcrowd 👉 https://itspm.ag/itspbgcweb
Devo 👉 https://itspm.ag/itspdvweb
Episode Introduction
Welcome to Redefining Society, the ITSP Magazine's podcast, where we delve deep into the intersection of humanity, society, and technology. I'm your host, Marco Ciappelli, on a mission to explore the impact of tech on every facet of our lives. From education to smart cities, cars, and IoT – the gamut is wide, and the conversations are intriguing.
Lately, many discussions have gravitated towards topics like generative AI, shared GPT, and DALE-E, sparking debates about copyrights and the broader implications of these technologies. Today, we're diving into something that's been a personal frustration of mine from my days in web development and marketing – the search engine.
While AI promises to revolutionize search, I've had my share of frustrating encounters with elusive search results. But today, we're exploring an exciting development that could redefine this landscape. Our guest is Eric Olson, co-founder of Consensus, a new search engine specializing in academic documents and research papers.
This podcast takes us on a journey through Eric's innovative vision, one that seeks to create a world where every question or debate is best served by empirical research. By leveraging the power of AI, Eric and his co-founder, Christian Salem, aim to sift through hundreds of millions of research papers, extract key insights, and present accurate and reliable answers. However, their approach is different – Consensus is about finding the answers, not creating them.
Beyond a discussion about search engines, this episode is an insightful exploration of the role of AI in shaping our society, the ongoing battle against misinformation, and the potential for technology to revolutionize our access to knowledge.
So strap in as we journey into the intricacies of how search engines, AI, and meticulous scientific research are combining to redefine our society. Remember to subscribe and share our podcast to continue being part of these captivating discussions. Extend your engagement with us by subscribing to my LinkedIn Newsletter and visiting MarcoCiappelli.com for more insights into the role of technology in reshaping our society.
Stay tuned, subscribe, and share – and remember, the future is now.
_____________________________
Resources
____________________________
To see and hear more Redefining Society stories on ITSPmagazine, visit:
https://www.itspmagazine.com/redefining-society-podcast
Watch the webcast version on-demand on YouTube: https://www.youtube.com/playlist?list=PLnYu0psdcllTUoWMGGQHlGVZA575VtGr9
Are you interested in sponsoring an ITSPmagazine Channel?
👉 https://www.itspmagazine.com/advertise-on-itspmagazine-podcast
Please note that this transcript was created using AI technology and may contain inaccuracies or deviations from the original audio file. The transcript is provided for informational purposes only and should not be relied upon as a substitute for the original recording as errors may exist. At this time we provide it “as it is” and we hope it can be useful for our audience.
_________________________________________
Marco Ciappelli [Host]: / All right, everybody. Welcome to another episode of ITSP Magazine's podcast, uh, Redefining Society, which is with me, Marco Ciappelli, and as you know, in this, uh, particular show, we dig deep into what's happening in our humanity. In our societies and, uh, all of that, how is affected, uh, by technology. So that's an easy conversation.
We can pretty much talk about any, any topic as you know, and I, and I like that. I like to talk about education. I like to talk about smart cities, cars, IOT, you name it. But lately, and I was joking with, uh, with my guests here. Uh, that I'm going to introduce in a few, in a few seconds. Um, one out of three conversation is probably around, uh, generative AI, shared GPT, DALE-E, and, uh, copyrights and, you know, whatever good and bad is doing in our society.
So this one I'm excited because when. The, the opportunity was presented to me. Um, I have been in the web development and marketing, um, before, uh, ITSP magazine and, uh, branding and search engine always being. You know, big part of advertising and marketing and communication and of course research and academic and, you know, you don't always find the results that you want, it's kind of like you throw, you toss a coin and you see, you see what's come back at you and it's been frustrating.
So, um, I think AI is helping with that, probably is going to help even more, but there is a actually a new search engine that is specialized into, uh, into more like paper research and academic documents from what I'm understanding. And so, uh, I was curious. To learn about this. And I don't think I can have a better person to talk about it than one of the founder, Eric Colson.
Welcome to the show.
Eric Olson [Guest]: Yeah. Thank you so much for having me, Marco. Yeah. Good to be here and talk about all things search AI and how it can impact us in the world.
Marco Ciappelli [Host]: Absolutely. I like to talk about past, present and future. So that's kind of like our script. But the first thing, let's talk about your past and your present.
So who is Eric? And how did you come up with the with this mission to change the search engine business? Yeah,
Eric Olson [Guest]: uh, so from the Boston area. Originally I come from a family of, uh, engineers and teachers and researchers. I was the jock in a family of a bunch of academics and I went to school out in Chicago at Northwestern where I played football.
Uh, it's actually where I met my co founder, uh, Christian Salem. Who, funny enough, has a very similar background to me that he is the only member of his family who does not have or is not currently pursuing his Ph. D. Uh, and we kind of bonded over that fact that we were these, uh, closet nerd jocks, uh, who loved and appreciated science and research and technology and love to talk about those type of things.
And we, you know, it was around the time I was graduating. And I believe it was really the first time I'd seen, you know, Googled featured snippets when they when you type in a question, it's able to pull you out a particular sentence from the document that effectively answers your question. I think it was around then I saw that technology and basically had the idea of.
Man, if that is possible, I couldn't theoretically be able to do that over really rigorous source material. And when we have all these questions and debates that could be best served, answered by empirical research, what if we could use technology to scan that research and analyze that research for us?
And give us consumable insights from the best content, uh, that exists in the world. And that was my idea like seven or eight years ago. I pitched to Christian. He thought it was an all right idea. Uh, we both worked in technology and started to. Gained some of the skills to actually execute on it. I worked in data science.
I was working at DraftKings, the sports betting company for about three to four years after graduation in the data science department. Christian was a product manager at a startup and then in the NFL media department. Uh, and then when COVID happened, uh, two things, uh, led us to start to want to work on this.
One was that, you know, misinformation was a thing back when I came up with this idea. But it had become an even bigger problem and it specifically was a problem around scientific topics in the heart of the, in the height of the pandemic. So it was this kind of problem statement that we had identified, uh, was very much in our faces at all times.
And we could clearly see there was some public demand for easy access to good, rigorous information. And then the other piece of it. Uh, was we started to see some of these seminal papers coming out about language models. Uh, and it was Christian who actually brought me the famous attention is all you need paper.
That was the impetus for this transformer architecture that has led to things like chat GPT. And this was back in 2020 when him and I were talking about it. So, you know, we're not the most language model hipsters, like some people who have been in it for a few years before that, but. Uh, we can say that we were doing it before, it was cool to be doing it before it became really mainstream and we basically said, holy crap, uh, this technology's gonna change the world.
And there's about to be a bunch of startups that are built using this technology that are gonna take off and we want to be working on it. And we thought that those two things combined, the problem statement, being so present and seeing the technology that could enable it being built, really taking off, uh, led us to start working on it, uh, in 2021.
And yeah, we quit our jobs in August of 2021 and started working on Consensus, which I should have done a better job of introing it at the start. But, uh, it is a search engine that is powered by language models that finds insights and answers in scientific research papers. So it allows you, regardless of your expertise level, to come and ask a plain English question.
You could say, is fish oil actually good for my heart? And instead of getting back, uh, you know, some health and wellness blog written by who knows what, uh, it will actually return you studies that have looked to answer that question and then pull out the insights from those studies. So you can see, uh, line by line of what each study has included.
And then we also will synthesize across the top X papers, uh, and get you an answer that is fully grounded in scientific research.
Marco Ciappelli [Host]: Yeah, that's pretty cool. So I've been playing. Quite a bit with, uh, with some of the API of, uh, chat GPT. So I, you know, I, I think, uh, and you tell me if I'm wrong or, or not, but it's kind of like working on a knowledge base that is not the entire internet or very generic, but actually focus on the archive where you do find.
Legit paper, legit academic research and all of that is kind of like more or less the vision here.
Eric Olson [Guest]: Not exactly. So what you're describing would be if we are to be building our own language model, they would be pre trained on this. And that's already been done. Galactica famously came out by Meta, which was a.
Tragedy PT like model that was trained on scientific research papers. And, uh, it was really bad. It said all sorts of really wonky things. And, uh, when you just are pre training it and then asking it to completely generate the text itself, like you have no way to understand, like the reasoning, the thought process and checking the work of how it came to these answers.
You're just saying, Hey, there's all this research mushed into this language model, and now it's going to generate me answers. Like that actually isn't that helpful. You really want to be able to, especially if you're doing research, Or professionally, like you really need to see the studies that these answers are being pulled from.
So what we've done is actually we've aggregated a database of hundreds of millions of research papers and we're basically using the technology on top of that. So we're using, got language model technology to surface papers and then pull out information from papers. So basically the simplest, this is.
This is dumbed down, but like, we go looking, we use language models to go looking for the answers as opposed to using language models to create the answers. Got it. So the example that you were talking about would be somebody who'd be trying to like. Build a model that can generate you answers from all of this knowledge stuffed into it, as opposed to having this database of knowledge and going into it and trying to pull it out.
Marco Ciappelli [Host]: Right, right, right, right. No, no, I'm glad that you actually, you know, air quotes for people listening and not watching, uh, down, but down and which is something that, you know, I don't see as a dispresitive thing, but my mission is to understand, understand things. And I know a lot of people are. Out there listening.
They are interested in this, and that's why they're listening and try to understand. And we don't need to go all geeky. Uh, you know,
Eric Olson [Guest]: everything in this world needs to be dumbed, down needs to dumbed needs to "dumbed down for everybody". No matter how versed you are. There are very few Uh tried and true experts at every single part of the things that we're talking.
Marco Ciappelli [Host]: Absolutely. Absolutely. So I like the idea that Did you that you can use examples for these and and I guess my question that is probably the one that people are Wondering at the time is how do you know that what you find is, is legit, right? I mean, we talk about misinformation, one of the fear about artificial intelligence, it's actually creating even more misinformation.
So it's always that adverse adversarial situation where yep, there's the good, there's the evil, it's a never ending battle. We use technology to solve technology problem and so forth. So how do you know that something is legit
Eric Olson [Guest]: Yeah, I mean, first, it's a problem that is always going to be being worked on and like, there isn't just some magic ball that says, Oh, we do this. That means that we now solve this information and everything we're going to show you is perfectly fact. Let's go outside and so it's like always going to be an ongoing problem.
There's, Uh, yeah, we've approached it like in a few ways. So, so one simplest one is, uh, the only content that we surface is from research papers. So like we don't index the entire web. We index. aggregations of scientific papers. So like we can't show you editorial as content. Doesn't mean that studies aren't fraudulent.
Sometimes studies can't be wrong. Studies can't have faults. So like still gets us in a better position, but there's still now is all this other stuff to worry about. Next one is kind of what I talked about before, that the way that we've architected it is it is pulling out insights from papers and it's all tied back to the papers themselves.
So it doesn't mean we can't pull out something from a paper that's A crappy paper, but you have the ability to check that yourself. We're not going to be making up information. We can't be hallucinating in the traditional sense that chat UBT could, where it's just generating you this text. We're going and looking for the information and pulling it back to you.
So that helps us a lot there. And then the last one is, uh, we do have like quality indicators throughout the product that try to say. With some objective standard, like, what are the green flags for this particular studies? Like, what type of design is it? Is this from a really well designed meta analysis?
It's looked across a bunch of studies, or is this just an end of one case report? So many more things we can do there, but continuing to add things that are telling you in an objective way. Like why you should care about this and why it's potentially higher quality research. It still doesn't solve everything, but there's a lot of things you can do on the individual source level.
So given that we found you a study and we pulled something out, we can run language models over that study the same way that a human expert would to assess its reliability. Who is it funded by? How big was the sample size? What is the design of the study? Did they do all these things? What journal does it come from?
We can use AI. Analyze all of those things and spit out metrics that say, like, here's where it ranks across all of those, meaning this is our confidence. And it's probably the right way to say it isn't going to say this is 100% true. This is 100% false, but we have ways to assess the quality of it.
Marco Ciappelli [Host]: Yeah, now that that makes sense.
So you got an algorithm that is asking the right question that You know, a human,
Eric Olson [Guest]: you automate what if you were to put the best scientists in the world to, uh, look at a paper and tell you if they thought it represented the truth or not, like you can train language models to take those same steps.
Marco Ciappelli [Host]: And actually, I want to throw this thing at you because I like to look at things from a philosophical perspective and then and then maybe we dig into more like case study.
But, you know, this idea that, uh, you know, people have. Fear of AI, of course, you know, copyrights taking job writers on strike right now for for the movie industry so far. So everything is legit. But I, I have this vision that AI is what you make of it, right? It's a tool. So any, and it's reflecting our humanity.
And I love when you said, even the biggest expert can't, can't be wrong. Wouldn't be the first time. We're not going to be the last time. So, you know, what's been the, the reception from, from the audience on this? Was it? I mean, I'm expecting when I saw it was like, finally, I gotta be honest, you know, I'm not paid to make promotion for these, but I'm like, this is good.
This is good because I don't want to hear what I don't look for. And it's great for advertising. Maybe we can touch on that. Don't, don't get me wrong. You know, if you like this, then you like that Amazon sponsor placement, that's how Google made all the money. But. I feel like should be welcome. So what's the feedback here?
I know you got some really good investment. So I guess that's that's the sign to it.
Eric Olson [Guest]: We were lucky enough to raise some money from a great group of investors. Recently, um, I'd say first we don't show any ads in the product and we want it to be. It's a subscription product and there'll be premium features behind paywalls.
We think that also is a way that we can help that last question you had of We don't have conflicts of interest where we're wanting to prioritize results over other ones. Our only incentive is to get you good information and it should stay that way. Um, as far as audience reception, I'd say it's, uh, it's been overwhelmingly positive, probably more than I even expected.
Uh, you know, when you are the one helping build this thing and understand all of the ins and outs and all of the warts and all of the shortcomings of it, you get pretty self conscious about that. Uh, and then you expect that everyone's going to come across those and have all these issues with it. And then you realize that that's just your like imposter syndrome kind of talking.
And when you actually get it out in the wild, um, how well it's been received has been just like truly humbling. Uh, we've grown incredibly quickly, really, uh, all organically. So I think that's the best marker of good audience perception is we've over 300, 000 registered users. We've done that in about six months with 0 in marketing.
Uh, and I think that speaks to the market readiness for a product like this, just like you're talking about, like you see all the, we'll see all the time of people tweeting about us and just saying like, where's this been all my life, or I wish this came out when I was in school. And, uh, yeah, it's clear people have thought about this problem.
I know we're not the first ones to think about this problem. Uh, and I think that combined with. It being just such a clear, positive use case for this technology, like makes people, when they see it, it just makes sense. Uh, and I think that's been a huge reason why we've grown so quickly. I mean, there's still haven't been critics.
There's still haven't been people who say like, you're, you know, you're, you're. Stepping around processes that really need to be done manually. And this is going to create this issue here. So like there's many of them are valid and they're things that we're trying to address and, uh, make better, but it has been more positive than I think I could have ever,
Marco Ciappelli [Host]: well, it goes go back to what you were saying.
You know, perfection and ideal is not made to ever be reached. I mean, it would be nice, but you know, but you need to strive to it. That's for sure. So I want to talk about this perception that. So the internet, I've been around for way before the internet. So, but when, when the internet come out at a certain point, it became like, if he's on the newspaper, it got to be true.
If he's on the radio, the radio said that the TV said that, then the internet said that, but the internet, it's, it's not editorial content filtered. So in a way for me, something like this, it's a, it's a way to, to filter what's going on and may not be perfect. But you, you know, you have an editorial review in this case by my, made by an algorithm and AI, but you know, still focus on, on that.
So you get to the point where people should pay for this. So I go back and it's like, well, if you make me pay for it, I'm going to go to somebody that gave me for free, then you got social media. I'm going to like, you know, condense social media problem in a, not in a sentence, which is of course, if you're not the, the customer you are, you're the product, right?
So the fact that there is a pain model, subscription model that the people are now accepting again, at least. From what I see, you know, it doesn't have to be free because people are learning this scrapping data Privacy all of that. So do you see do you think this is the future of? The internet or the web 3.
0, whatever it is where you know, you want a good service. You're going to good product
Eric Olson [Guest]: Yeah, you gotta pay for it. I It's a great question I, I don't, I would not say it's the future in the sense that it will be all encompassing, but I think it will be a significantly larger chunk of the tools that we use will be subscription based on what we're used to.
And I think it will be because of two reasons. I think one, it's exactly what you're saying. Like, there is this. Much larger understanding that if you're not paying for it, there's likely some things that you're not too happy about going on behind the scenes, like not paying for the product, you are the product, uh, and people like really deeply understand that now to a level I don't think anybody could have realized that people were going to have this, uh, giant understanding of that.
It was not true. And all of these platforms were being launched. Decade, two decades ago. Uh, and then the other one is, is the technology itself that it's getting to the point where it can be so good and can automate so many parts of different processes that people will be willing to pay for it, that it actually will produce that, that magic that gets people to pay for it.
But I don't think it's right for every single use case. Like there still will be. Pure consumer facing tools that we'll figure out the right way to put in advertising alongside language model technology and search products. Like there's no doubt that will happen to some extent how exactly they do it. I don't know.
That's not my problem. Uh, we'll let Google figure out how to do that with Bard or Microsoft figure out how to do that with Bing and integrate the two into a seamless way, because they're probably just pushing that aside and saying, we'll worry about that later while they try to, while they arms race each other on the technology.
Um, but yeah, I think. For our sake, like it is, it's really important for the core product. Like it keeps our incentives aligned with the, with the user. And then we have the benefit that there's so many people using our product professionally and in school. And that gives us the ability for people to pay via that mechanism.
So people to have their company expensive to the company or expensive their university or have their university pay for licenses for everybody within the, uh, within the school and what we want to put behind paywalls. Are those features that people who are really engaged users are using it for? So we are going to keep many, much of the search functionality free, where if you want to just pop in, see if that supplement you heard is legit or not, you can do that and just get a search, get the results back.
But if you want to bookmark a bunch of papers or do a deeper analysis where you have a GPT 4 co pilot alongside with you into a paper, like that will then be put behind the paywall. So, Paywall things that people are using, who are using the product in a really engaged way and getting tons and tons of value for, then it's, you know, it's a, it's a fair trade effectively, right?
Like you're paying us cause you're using this product so deeply and getting all this value out of it. Whereas then we can keep other parts of it free where it's just people popping in once or twice. And I think there's going to be so many products that do things similar to that. Uh, the best, my fit, I shouldn't say the best, my favorite.
Uh, search startup that is not in our space. The, the, my favorite generalist search product. We're like a vertical search. My favorite generalist search startup is perplexity, perplexity AI. And they're doing a cool thing where their search functionality is free, but then you can enable. Uh, this like cope GPT for co pilot alongside your search results so you can do all these cool things with, and then you have to pay for that.
I think we're going to see lots of search engines and other products using our ones do similar things. We have some parts of it free, but then you have these deeper integrations and different features that will be subscription based. Sorry, very long answer, but no,
Marco Ciappelli [Host]: it's great. I mean, I won't long answer because then I want to ask you something.
I'm like, not just change my mind because he said something else. I want to ask you something else. So the last one that popped in my head. Connected to what you said, and I love your perspective on this is we're going from because technology allowed it to generalize in like, I don't know. I'm going to go back.
I'm going to date myself again. If people are old enough like me, they remember AltaVista. They remember all this old search engine and they all that Yahoo and they will become a portal. So it's like we were born like a search engine, but then we're going to give you the email. We're going to give you the news.
We're going to give you the music. We're going to give you whatever the hell you want, because we want you to come back here. We need those eyeballs. We need those clicks. That's how we get advertiser. Blah, blah, blah, blah, blah. Then, uh, Google came in and it's like, no, we're just going to do search because.
Found a way to make money out of it, but it's a search about everything, right? It's almost like 42 the answer to to everything in the universe and the meaning of life and you guys are going much more Focus on certain specific things. So the question is for you We reached to the point that it just doesn't make sense to know everything about everything, but we need to be specialized in and this is not just for search engine.
I'm thinking in general, like the Acme, the company that makes everything. I don't think it makes anything well Ask Willy Coyote Uh,
Eric Olson [Guest]: I
definitely agree with you to some extent that was going to be part of my answer though Yeah, it's like our take on it is A generalist product that is trying to solve millions and billions of different types of problems for people will never be as good at Uh, as us when we're, we're focused on a specific subsets of the type of question and topics, but that only, in our opinion, that only matters for certain domains and topics.
And that's why we are excited about what we're working on because we think that. Science is one where that is the case, where you really do need a specialized focus on it to actually address it properly. But I don't think that's the case for everything, and that's everything else that, you know, can be done with a generalist product and doesn't require this, like, immense focus and specific purpose built features for.
Let that be solved by generalist search. So like, uh, an example is like, uh, like travel planning or something like, I don't think I'd really want to build a travel planning search engine co pilot product right now. Like that's probably just going to be figured out by whatever generalist search product AI product becomes the best, or like a recipe builder.
Like those are things that like can just be done with whatever wins the, the general consumer search consumer search battle, but then there are other verticals. Where you really need to go deeper in order to do it well. And we believe that science and research is one of those vertically. Uh, so like we will always have the advantage of being able to build these special purpose Workflow related features and ones that go deeper that like you just would never build if you're not focused on it
Marco Ciappelli [Host]: So people listening and the dogs maybe the mic is picking that up. But again I'm not going to edit this. So those are my dogs. Um, so is there something that, that is interesting to me, which is you, there are certain vertical that works on your opinion better than other. Maybe somebody thinks, well, there is an opportunity somewhere.
Eric Olson [Guest]: I'm happy to be wrong. And there's definitely things I'm not thinking. No, no, no. Absolutely.
Marco Ciappelli [Host]: I mean, you, you don't know what's going out to, you know, what's going to happen tomorrow, to be honest with you. But I'm wondering if, Um, if, and this is putting our futuristic hat on right in the, in the future, which we're living already in it, AI, uh, generative, um, AI will be able to do this filter.
On his own. And that's a question mark. So like you, you have focused on this, but what if you say, I'm imagining I go to chat GPT four and I say, you know, connected to being which now you can do and say, all right, I'm interested in this vertical. I don't want to hear any other. Do you think it will be possible or not?
And I'm not, I'm not saying this to what you're doing. It's just a legit question. Like, You know, to what extent AI could be able to do this on its own or not?
Eric Olson [Guest]: It's a fantastic question. And if you are not asking yourself that question, if you're building an AI startup right now, you, you're wrong. You have to be asking yourself that question.
The reality is nobody knows what, what the answer is. It's like basically to dumb down the question. It's just like, will generalist AI just solve every problem that startups traditionally have solved alongside big players? I think this should be a question on legitimately anybody who is working in building in the space right now.
If the trend of any other technology is true. It won't be the case, but you can make the argument. This is a type of technology that we've never seen before. I still believe, I think two things. One, it is unclear, uh, how much more progress there is to be had in the next few years. Like we don't really know the status of GPT 5.
Sam Altman alluded to that, uh, it might not even actually be trained in right now. So we might be years away from that. And. You know, there's a chance that that just is wider and does more things. It doesn't necessarily allow you to go that much deeper. We, nobody really knows exactly what the net of that will be.
There have been other technologies that have kind of hit ceilings for periods of time, and we may hit one with this. I don't know. Maybe we don't. But then the other thing I think is, so that's one case for why it wouldn't that like. In the next 10 years, we don't really progress it like that, that much further.
And I think a lot of people in this space are interested in instead of just making bigger and bigger, bigger, making smaller and specialized and more efficient and able to put on different devices and more production settings is like a lot of the focus in the space right now, which is making it more practical as opposed to blowing it up and having to solve every problem.
So we might be further away from that than people. Uh, you know, especially a few months ago when hype was at its peak, might've believed again, still, if you also are not saying I'm willing to update my opinions always in the space, you're also wrong because everything changes all the time and trying to break through stuff all the time.
Uh, the other piece that's an argument for why that wouldn't happen is I think people underestimate how much, like we reset what is a baseline. And what I mean by that is the difference between something that is good and great, like. It may seem like that difference isn't that much. But then when good becomes just your baseline, that now that relative difference is massive.
And I think that will always just keep happening with technology that even if generalist, like the example of a company, like a AI startup right now, that by everyone's account is. Completely done away with because of generalist AI is like a copywriting tools. Like that was one of the first things we saw built.
General AI is, Hey, write me a piece of marketing copy for my website. And then you can just do that for you. But if they just get like a little bit better and like. Everyone's baseline is now chat GPT. Is that difference between now the baseline and there's being 10% better. Does that now the difference what used to be between zero and some percentage, right?
Like I think people reset what is baseline more than you realize. And those small differences may become big relative differences. Uh, and I think that process will probably just keep happening. And yeah, I, I lean toward generalist, not solving everything in the next few years because of those two reasons.
But, happy to be around.
Marco Ciappelli [Host]: I, uh, I'm going to agree with you. You know, I mean, I am concerned for the mediocre is going to be swollen by that.
Eric Olson [Guest]: That's exactly kind of my point. Yeah.
Marco Ciappelli [Host]: If you're, and I don't want to offend anyone, but you know, a mediocre artist may be worried. I mean, Dr. Ryder could be worried. I have seen stuff that I would not use.
I use it for ideas, but I usually don't just take whatever it gives me and just commoditizes
Eric Olson [Guest]: mediocre work in general,
Marco Ciappelli [Host]: again, research, uh, ideas, creativity, whatever you want, but then, uh, if you think that it's great. I think you're wrong. I think it's okay. Right. So,
Eric Olson [Guest]: but what is it now? Now our baseline becomes okay.
And then the startups have to build to really, really great to make that relative difference enough for you. Exactly.
Marco Ciappelli [Host]: It's almost like it would be better. It pushed us to make even better, more human, more creative
Eric Olson [Guest]: things. It will push Jasper AI to make their marketing copy converted. X percentage that then makes it worth it to pay for versus what you get for free on your generalist search product.
Like, I think that will probably just keep happening.
Marco Ciappelli [Host]: Well, you know what? We're going to leave all this latest questions that we raise for the audience to digest. And, and as I always say, if we leave, uh, If we finish an episode and people leave us with more questions than when they started to listen to the show, I'm happy.
You know, I want people to think, and if they think, and now they actually play instead of just criticizing, because that's another thing, many times they just criticize what's happening. You know, there's always an early adopter and... People that don't believe in it. And then, you know, people that say, Hey, I think now I can use it because somebody already tested it.
But, um, I, I was excited to learn about what you guys are doing. And it gave us the opportunity, of course, to talk about a lot of different scenarios and past present and future. So Eric, thank you so much for being part of the show. And, uh, I want to remind everybody that, uh, they should share this, uh, conversation, subscribe to ITSP Magazine and in particular to my podcast, Redefining Society.
And, uh, and again, uh, everything, links to, uh, Consensus and, uh, social media for Eric to get in touch with him and co founders and everybody working on this will be in the note that I can tell you. It will be written by Chat GPT. I'm not afraid. So I'll give you some
Eric Olson [Guest]: when you when you can give it a template and say, Hey, it's perfect for that.
Marco Ciappelli [Host]: But I know that we have this conversation. The two of us are pretty sure we're real. So that's that's good. Thank you so much, Eric.
Eric Olson [Guest]: I really appreciate it. It was really fun. Thank you so much. All right.
Marco Ciappelli [Host]: Take care. See ya.