In this second episode of our “Generative AI” series, Cornell Tech and SC Johnson College of Business professor Karan Girotra joins us once again to assess the current capabilities and business uses of generative AI tech and to examine what's coming next — as well as what's not.
Cornell Tech and SC Johnson College of Business professor Karan Girotra joins the Cornell Keynotes podcast to explore what’s new in the world of AI, including updates on Apple Intelligence, Anthropic and advancements in China. We examine late-breaking technical advances in generative AI such as new video capabilities, autonomous agents, robotics and the next generation of models.
The Cornell Keynotes podcast is brought to you by eCornell, which offers more than 200 online certificate programs to help professionals advance their careers and organizations. Karan Girotra is an author of three online programs:
Learn more about all of our generative AI certificate programs.
Follow Girotra on LinkedIn and X, and register to attend his upcoming AI Today Cornell Keynote.
Chris Wofford: Today on Cornell Keynotes, we continue our monthly series on generative AI with Cornell Tech and SC Johnson College of Business Professor, Karan Girotra. Karan reveals what's new and hot in AI tech, including a first look at Apple intelligence. We review Anthropics Claude and look at eye popping advancements coming out of China.
Chris Wofford: We'll also offer insight into how organizations are transforming their strategies through generative AI and how you can envision your organization's path forward and start executing. Be sure to check the episode notes for details on eCornell's multiple AI courses and certificate programs authored by professor Girotra. Listeners, here's my conversation with Karan.
Chris Wofford: So let's start with big tech. What's going on in this space right now?
Karan Girotra: I think since we talked last, it's been about a month, and if I had to think about what changed, what's new, I think there are three kind of notable things that I, I have registered for myself. First, the big one, the big announcement we had over the last month, month and a half, was Apple Intelligence.
Karan Girotra: For a long time, we didn't know what Apple was going to do about AI. They showed us their product. Second, somewhat, somewhat notable thing which is coming from, quote unquote, one of the big tech companies building AI is from Anthropic. Now, Anthropic is, is not a company you may have heard of. It's the company started by some former open AI engineers, backed mostly by Amazon.
Karan Girotra: They put out a new model which has which has got some attention for some distinctive features. So, it's the second thing to talk about. And then a third thing which, which is my own kind of personal thing. You won't hear about it too much on Twitter and other places is what's been going on in the world of AI in China.
Karan Girotra: We've been very U.S. focused and as our viewers must know, China and the U.S. Have become somewhat distinct ecosystems as far as technology goes because of all the geopolitical reasons and while the U.S. I think is still leading. There was some notable announcements from China, and I think we should keep our eyes on what's going on there, too.
Karan Girotra: So I think three big big things, Apple, Anthropic, and then really what's going on in China. I think these are the three big things we should, probably unpack a bit today.
Chris Wofford: So let's start with Apple Intelligence. Give us, give us a little preview on how the tech works, what the user experience is like, and maybe say a thing about what it does for business, right? And Apple's competitive either advantage or disadvantage in this space, relative newcomers, right?
Karan Girotra: Yeah. So I think as, as Apple always does, they were not the first out of the gate. Okay. But their goal is to be the best out of the gate for the particular segment they focus on. And now, let's keep in mind, Apple is very much about individual productivity.
Karan Girotra: How can users, business users, individual users, or consumers use AI to make their lives better? Apple is not really an enterprise technology company, even though they've tried spinning this technology here and there, their products, but by and large, they're a user, consumer focused company. And if I had to describe what Apple intelligence has put out in the context for its users for its target market, which is, individual users in one word, I think the distinctive feature here is, "seamless integration". Unpack that. So technically what Apple has put out is not that different from what you could do using chat GPT or the other products from open AI or other, other big AI companies. You still have the basic writing and summarizing text kind of features that you can do with chat GPT.
Karan Girotra: You still have the features which can generate images in different styles, emojis, real life illustrations. We've been able to do that with MidJourney DallE. And third, I think they have an LLM agent type type feature in there, which is also something which is doable. Now recall agents are, are a thing where the large language model creates a recipe for a task for you to do, and then it coordinates other programs to do that task for you. So it kind of takes you what you want to be done and breaks it down into small tasks and then communicates with individual applications to do those tasks. So technically, the three big pieces are the same as what we already had.
Karan Girotra: Generating text, generating images, and having these large language model agents which can do tasks for you. But what is different? It's what started with seamless integration, all of these features will be in your apps. So the text emoji, it doesn't mean you have got to open chat GPT now to write a great text message, then copy it over to messages.
Karan Girotra: No, it'll be inside Messages. Anywhere you write, you can get writing help. If you want to generate an emoji, it'll be available within the if you want to generate a message or visual on a text message chain, it'll know that you probably want to send an emoji, and it will be available right there in the app.
Karan Girotra: So it will be everywhere in a way, but visibly nowhere. Right now our model of AI is it's an application. We use it or it's a website. We go and then we copy, paste and try to do stuff with it. Apple is bringing it within the ecosystem, within every, every interface that you use. And Apple is probably the only company that can do that because they really have control over the ecosystem.
Karan Girotra: So Apple's AI technically won't be much more sophisticated than, than everyone else. But it will be everywhere, and in terms of at least performance or things it can do, it'll be very comparable to others, but it'll be everywhere, but visibly nowhere. It'll be seamlessly integrated to create a beautiful user experience.
Chris Wofford: And I think the differentiators are what stand out for you, right? So, what is it about Apple's products? I think I know the answer, but I find your particular slant really compelling, so tell us what differentiates Apple in your estimation.
Karan Girotra: Apple Intelligence, which is what they're calling their AI. So Apple intelligence will be different than all other AI in the same way as your iPhone is different from every other phone.
Karan Girotra: It's going to have a magical user experience. You'll need fewer clicks to make the stuff happen. You won't have to copy paste, deal with all the wonky clunky stuff that you have to do. Think Apple TV versus watching something on Netflix. Apple TV is front and like if you watch the Apple Plus shows where you fire up your Apple TV, it's right there.
Karan Girotra: You press one click and you can start watching your TV. You gotta go into Netflix, you gotta log in, you gotta choose the user, blah, blah, blah. It takes three more clicks. Now that might not sound much, but reducing user friction is, is where Apple excels. Making it magical, making it AI for everyone. So I think the experience will be very good.
Karan Girotra: And I think we almost saw that. Microsoft is another company which has which has a ecosystem of products beyond AI and the demos we saw from Microsoft and then we saw some Apple's demos. Frankly, I think Apple kind of was a lot better. MS was wonky. It was kind of confusing how you'd use it like many other Microsoft products and Apple was like, you don't need to know how to use it. It'll just be there. It'll be natural. So I think that that is definitely going to be there. And I was pretty impressed with the announcement from Apple mostly because, because of the user experience, and then there is a technical kind of difference that Apple is bringing in there.
Karan Girotra: And those technical differences manifest them in again, a thing that Apple cares a lot about: safety and privacy. Particularly privacy they went all in on and in particular, protecting the customer's information, they've really made some technical differences, which I think changed the game for customers in how to think about privacy of using AI products.
Chris Wofford: You know, and privacy and security are something that we've come to rely on over the decades with Apple products. Some of us complain that it's a little too, it's a little clunky, it's a little much, right? But in the end, we are enjoying a great deal of privacy. And you say there are multiple levels as it relates to Apple Intelligence.
Chris Wofford: What are we talking about here? How's it going to be different?
Karan Girotra: Yeah, I think, see, first thing I'll say is the context in which AI is being built is very different than the context in which social media was built 15 years back or so. I think today, 15 years back, people trusted tech. Today, I think I talk to a lot of companies, big companies, small companies, individuals.
Karan Girotra: And despite all the stuff Microsoft and others are talking about it, privacy remains the number one concern. For people who've gone beyond, who found some utility in, in AI, they're like, oh, and what's going to happen to my data? So the world is different today than it was when the previous generation of technologies came around.
Karan Girotra: So I think privacy is currently one of the biggest problems that we need to, at least in the consumer's mind address. That said now, what is Apple doing in here? Let's think about what the whole whole landscape of privacy is from the other offerings and then contrast Apple with that.
Karan Girotra: First up, any of the big AI companies, if you're using a paid version, if you're using a paid version of ChatGPT, all of them offer versions where you have to pay something, but with those versions, they promise that we will not use your data in any way that you don't permit us to use it. The important word here is promise.
Karan Girotra: All of these products, be it Chat GPTs, professional services or Microsoft Copilot or every other kind of AI company. They legally contracts themselves not to do it and they say, trust us, believe us, we're going to promise you. Now, like I was saying, this is not 2010 or 2012 when we believe the promises of big tech companies.
Karan Girotra: So Apple has gone further. Apple says, you don't have to trust us. Trust us. The way we will implement our AI, it'll be technically impossible for us to use your data in ways that you do not permit. So big difference. One is like, we could do it, but we're not going to do it. Trust us. We're not going to do it if you pay us an appropriate amount.
Karan Girotra: And we have this long agreement, which you hopefully have read, and then trust us, we're going to follow this agreement. That's what everyone else is doing. Apple is doing, you don't need to trust us. The way we are designing it, it's going to be technically impossible for us to use the data.
Karan Girotra: Now, what is this technical design? Let's understand how can they make that kind of you don't even need to trust us. It is just impossible for us to do. And I think the core idea here is that you use, so when OpenAI puts out an AI product, they're putting it out in a way that you always get the best answer and ignoring all kind of privacy risks in there.
Karan Girotra: And I say risks because it's not like they're going to steal the data, but oh, the pipeline might go somewhere where it might leak because of some reason we did not foresee. Apple has a different calculus. They're like, yeah, we want to give you the best answer, but we also want to take the level of leakage risk, privacy risk, only the level that is needed for that task.
Karan Girotra: And I think the dirty secret of AI is there's a lot of easy tasks we can do with AI. Those don't need the fanciest models, which need to be computed on the cloud, which is where a lot of the privacy risks start happening. So Apple's main idea is technically we're not going to send every AI request to the same large, big, complicated model.
Karan Girotra: We're going to route it. Keeping in mind the need for the performance and the potential risks that come from using more complicated models. How does it manifest? How does this principle of taking only the amount of risk you need for the task at hand manifest itself? First, a lot of the things you'll ask Apple Intelligence to do will be done locally.
Karan Girotra: That means the data never leaves your device. And Apple has made some good investments with these M1 chips or the, like the newer chips M1, M2, M3, and I believe M4 chips now, so, they can actually do a lot of this stuff locally. This is the best you could do. The data hasn't left your device. So in theory, there is no one else who should have access to it, but the local, the computing bar in your device is not sufficient.
Karan Girotra: So sometimes we do need to send it to the cloud, to bigger computers in data centers to process these requests. When the Cloud is needed, Apple is going to first send these requests to what they call their private cloud. This is think of it roughly like this is a private area on the server, which is going to be leased to you.
Karan Girotra: And they're using the latest technology in cryptography and encryption. To make sure that it's a little bit private part of the compute that you get to use when you need it, and then you're off it. So in a way, it's like you're getting your own little island, which is protected. And they're using, to the best of my knowledge, to the best of my colleagues, who are real security experts, tell me that this is a state of the art in terms of using cloud computing on a private cloud.
Karan Girotra: And then they have a third level. When something cannot be done on their private cloud, cannot be done in the local machine, then yes, they will send it to a frontier model, they will send it to the best in class models, like the ones from OpenAI. And, and when they do that, they'll ask you, can we send it to OpenAI?
Karan Girotra: So it'll be completely opt in. So overall, I think they're getting to privacy. The big, big idea is: youdon't need to trust, we technically can't do it. And the key way to make that happen is to not always use the most risky model or the most capable and most risk risk exposed model unless it is needed. So first try doing stuff locally, then in the private cloud, and only when where it's absolutely needed.
Karan Girotra: Kind of send it off to someone else who in the cloud, in these open AI type clouds, which have the most, where the pipeline has the most risks of being exposed, more information has the highest risk of being exposed.
Chris Wofford: This is a really compelling business model. How does it strike you? I know we're in early phases here, but what do you think about how they've done it?
Chris Wofford: Is this kind of a, a classic Apple thing? I mean, this is a, It's the same lane they've always ridden in some ways, but they are also doing something wholly unique within the industry, no?
Karan Girotra: So exactly that. So I think Apple is playing to its strengths. Apple, if you had to define the entirety of what the company has done, it's always been about integration.
Karan Girotra: It was between hardware and software. It's between your AirPods and your and your iPhone and your Mac. So Apple is all about integration, and they're bringing that same idea to AI. Apple is all about we charge the customers for the product. It's a products company. It's not a products and services company that individuals pay for.
Karan Girotra: It's not an advertising company. So they're leaning into privacy because it works with their business model. So, and I think they're also going to take a different track on safety. Different than privacy. Safety means what kind of responses you can get. Could you get adult content on these on these devices or using this AI?
Karan Girotra: So I think again, like they've done with these devices, they're going to keep it up a nicely controlled ecosystem. So I think overall, Apple has done. I was impressed. I was impressed, not by the necessarily the technology, but how Apple it was. It really was, I think, a company which knows their strengths and is playing on that.
Karan Girotra: Now that said, execution is still hard. And Apple has made substantial, it's not like Apple always executes correctly. We know Apple Maps didn't work so well. Apple tried social media products that didn't work so well. So it's not a given that this strategy will work. It also is not clear what customers want.
Karan Girotra: Maybe customers want or don't care that much about privacy and they want the, want the kind of best, best in class model all the time. So we don't exactly know how it'll work, but, I would say good start. Very good start. If I had to predict, I would like, the stock markets did. Apple had, I think, an 8-10 percent kind of a pretty substantial bump by showing this product out.
Karan Girotra: I'd also bet on them to get it right. I would just say it might not get right in September or the first version that comes out. They might take a couple of iterations, but eventually I think they will, this is important enough that they will get right. So think Apple Maps. They're not gonna, they didn't get it right initially, but eventually they did get it.
Karan Girotra: And by most measures, it's a pretty, pretty competitive product with Google Maps today. So I think, yeah, I'm impressed, and I'm cautiously optimistic that this will turn out to be AI for everyone. Now, I do want to make a distinction. At the individual level, I think we will, we will all, many of us will like Apple AI and use it, maybe in a couple of years, not right away, assuming they can execute and keep up with the technology.
Karan Girotra: On the enterprise side, I think Apple has never really been been able to kind of pull this off as nicely as uh, nobody's been able to pull off beautiful customer experiences in the enterprise. And at some level, these things matter less in the enterprise, at least that's the world we are stuck in.
Karan Girotra: So I think on the individual side, I'd bet on them. On the enterprise side, I think yeah, we don't know what, what exactly will happen. I wouldn't, I, yeah, I'm much less confident of Apple being able to bring AI to the enterprise, but to the individual, definitely, I think I feel good about what they put out.
Chris Wofford: Great. So let's move into a big tech story. Number two. We've got a company called Anthropic, who has a cool new solution for project management. It's called Claude. It's a collaboration feature where you drop things like documents or code in, and it helps answer questions related to a project. It builds things as projects, which of course is a collaborative endeavor, right?
Chris Wofford: So helps with workflows, productivity across teams, and it can do things like generate code or a text document. What's so special about this? Anthropic's Claude?
Karan Girotra: Again, in terms of the technology, not that different from what OpenAI is. Now that is, that is notable that it is, it's not different, but it's not significantly worse.
Karan Girotra: So the good news on the technology side is whatever Claude is putting out, technically, it's pretty close to what OpenAI has, and OpenAI still remains maybe the leader, but the margin has become almost negligible, particularly between, between what Claude is, Claude Sonnet 3.5, I think, what they've put out is very close to what, what OpenAI is GPT 4.0 looks like. So that's, that's good. you're not behind on the technology, but they're trying to, so how do you differentiate? You're not ahead. But you're not behind. So, okay, you're, you're even, but OpenAI has a two year head start. Everybody's heard of OpenAI. Very few people have heard of Anthropic.
Karan Girotra: So how do you, how do you differentiate? I think they're doing a few things. One, they're trying to make the model a bit cheaper. I don't think that's going to be very sustainable because OpenAI can, in the end, everybody's buying the same chips. So it's not, not that you can really have a big cost advantage.
Karan Girotra: What they're differentiating on is exactly what you kind of mentioned. I think it's mostly an interface that is different. So, all of these programs, ChatGPT Claude from Anthropic, and all of these programs can generate artifacts now. Artifacts might be text, might be pictures, might be code.
Karan Girotra: Others don't call it artifacts. Claude comes in and says, Oh, everything we generate, the output you create from that conversation is what we're going to call an artifact. And what they've made a little bit different in the interface is ChatGPT's interface is conversational. I say something and in that same linear flow, it gives me the code and then it says what's your next, next thing.
Karan Girotra: Simply put, Claude's product has two windows, one in which you're talking with the agent, in the other it's putting out its artifact, whatever it's generating, the picture, the code, and in that window, that window is an active window, it can sometimes run the code, it can compile the code for you, it can show you the output, you can go in there and tinker it with it directly, so I think the interface is now, is not, oh, I'm talking to the assistant in one linear chat, it's not chat, it's a chat along with a workspace.
Karan Girotra: Workspace where you're collaboratively working. Now, at the first level, this is a workspace for you and the AI to work together. But you can see this extends next level also. Different individuals working in a team can share that workspace. They can see the output together. They can be common files that they can use.
Karan Girotra: So I think without getting into too many details of exactly what, interface changes they have, the big idea here is they're making it from a linear chat interface, to a chat plus a common workspace type interface. I think how good or how bad it would be. What I can say so far is people who use, kind of, I would say the early adopters who use AI to write code, et cetera, are, enjoying this.
Karan Girotra: If you follow the vibes on Twitter um, Anthropic has pretty good vibes. Do regular folks care enough about it to switch from an OpenAI product to this? I'm not sure, but commendable effort. You're coming up with something new. And probably your only chance because the technology is getting close enough and everyone else is ahead of you or at least came out sooner.
Karan Girotra: So at this point, I think interface integration, these are the angles people are playing on and, Anthropic is doing the interface idea to make it more usable.
Chris Wofford: Good. Well, we'll follow the Anthropics Claude evolution and iteration and versions that come out because I think there's, maybe more ahead that's, that's certainly interesting.
Chris Wofford: Okay, finally, what's happening in big tech in China?
Karan Girotra: Yeah, and before I get to that, I do want to say one thing. I did say a few times OpenAI is ahead. But I don't think this is a done deal. I think i'm talking to a lot of companies and I think there's still a lot of potential. This is not the end of the story. If five years or 10 years from now the best AI products that or the AI products that are that are earning the highest revenues are from Open AI I'd be a bit surprised.
Karan Girotra: I think we're still in early stage. This is the Alta Vista of search. I don't think Google search anybody has kind of designed right now. It might be OpenAI which designs the kind of killer product, or it might be still anyone else. We're still in the first innings of kind of who's going to win the battle on the tech side.
Karan Girotra: And I think we haven't even started the game on the business side, which, maybe we'll talk about in a bit. Sorry, I didn't answer your question right away. I think in the U.S. we've been focused quite a bit on what is happening within OpenAI and sometimes all the drama associated with this company.
Karan Girotra: But let's not forget that the Chinese big companies Alibaba and others, have always been, have a lot of talent, have a lot of compute, and they've always been quite, quite good with old fashioned AI. Now, what do I mean with old fashioned AI? This was image classification, image recognition, driving assistant, on those kind of applications, which is what we used to call AI before the new GPT, Chat GPT, and generation of text and images came out.
Karan Girotra: So, in the old fashioned AI world, Chinese companies were already kind of leaders, for example, think DJI, it's a company which makes drones, I think they're best in class kind of capabilities to for these drones to fly themselves. EVs in China have pretty awesome driving assistance features, I think comparable and some would argue better, a little bit darker, but the facial recognition and security type AI.
Karan Girotra: China has always been ahead of U.S. companies for a variety of reasons. So I think China has this very strong kind of base, and now we're seeing some generative AI products which are coming out, which is pretty impressive. The gap between, I think, the leading models from China and Open AI is not as large as people might think.
Karan Girotra: So I was checking out Alibaba's product called Qwen. It's really good. It is in most benchmarks. It's comparable to what anyone except open AI has. It's comparable to Meta and others models, which were quite impressive. It's a GPT four class model on the video side.
Karan Girotra: OpenAI has shown us SORA, but it hasn't really allowed us to play with it. I think to the best of my knowledge, which is the only company which is making video generation easily accessible to consumers is a Chinese company, which has, I'm not sure I can say this right, but Kuaishou has a, it's a TikTok like company, but the not TikTok, the other leader in short form video in China.
Karan Girotra: And they, in their product right now, allow you to generate video, which I don't think any U.S. company is doing so far. So I think all I'll say is keep your eyes out on what is happening in China. It's covered less in the U.S. media, but I think both for businesses and individuals. It's important to see what's going on there, both to learn from it, and it's, possible that China will catch up and lead the models coming out in the U.S.
Chris Wofford: Great. Terrific tech update. So I want to ask you, I'm thinking about our viewers now, right? So all useful, what are you hearing around AI and what organizations are doing for business transformation. So I'm thinking about our audience here.
Chris Wofford: People who are at businesses of all sizes, right? Different objectives, they're at different phases of sophistication in their AI journey, maybe have not even embarked on them. Where do we start? Where do you think the greatest opportunities are?
Karan Girotra: Yeah, I think to answer that question, we got to look differently at different audiences.
Karan Girotra: So there's big companies, there is small and medium businesses, and then there is individuals or people who want to advance in their careers. I guess I'll start with big companies because they are the ones who started doing AI first. A lot of talk, company after company, I think you almost couldn't get away without putting out a press release on how you're going to use AI.
Karan Girotra: If you're a Fortune 500 company, you needed to talk to your board about what you're doing in AI, not really talking AI strategy, somewhat interestingly, but Oh, what are you doing around AI? most companies put out like, Oh, we're going to put we've created a chat bot for our for our employees, or we're talking to open AI to do stuff.
Karan Girotra: Good responses. But to be honest, Chris, I'm a little bit concerned about what has happened. Good energy, but I think this energy has so far manifested itself in small potatoes and some might argue even gimmicky stuff. Let me explain. The way I think about AI, it's a general purpose technology and general purpose technologies have several levels of transformation they bring to businesses.
Karan Girotra: First is productivity at the individual level and then at the organizational level. Third, individual productivity, organizational productivity. The next level is really new business models that are enabled by AI. And the fourth model would be new competitive positionings and new organization of the industry who becomes the leader who becomes behind.
Karan Girotra: If you look at companies, they're mostly looking at one third of the first level, the first level being just how can we use it to make employees more productive and even for employee productivity, there are many ways of using AI as an assistant as a co intelligence, which augments your intelligence as a companion as a coach.
Karan Girotra: Most of what you're seeing out there is just the assistant use case. So I think what most companies are doing, is OpenAI produced a consumer app, kind of an assistant, and then making it kind of available with some modifications to individuals. It's great, but it's one third of the first level of the transformation that comes out from AI.
Karan Girotra: That in itself is not a problem. You've got to start somewhere. But what we're seeing even in the implementation of this first one third, and it's probably got to do with the urgency and the hype that got created around AI, the way people are implementing this small piece of how AI can transform itself, people are seeing some mixed return on investment, mixed ROI. It's not unambiguously as good as we were hoping it would be. Now, why is that? This is something that bothers me. I've looked into it and I think it comes down to kind of classic mistake people make with technology. People who implement technology in businesses will either know a business very well or know the technology very well.
Karan Girotra: The truth is you got to really understand technology capabilities and your business's needs. Or your industry's kind of structure, the ecosystem, employees, how, what, what kind of cloth are they made out of, how do they use stuff, you've got to know those details. and so you need the contextual details, but you've also got to understand technology.
Karan Girotra: You can't take technology's terminator, some magical thing that'll come help you do everything. And I think that's what's missing. We're starting out with only the, we're doing only a little bit of what can be done. And some companies are not observing great outcomes from it, because they're not bringing the marriage of technology and business as well as they could.
Karan Girotra: And as a result, some people like, ah, this AI stuff doesn't really work as well. And we're getting that kind of hype cycle. Oh, it's so awesome. It's going to change everything. We tried something and now it's not really working. And I'm afraid. They're going to give up without fixing the two things that matter, really thinking of different levels of AI transformation and second, integrating business and technology.
Karan Girotra: I couldn't say this. I've repeated myself a few times already, but I couldn't say this enough. Do AI, great we got these great starts, but you got to do it right to get the full benefits and doing it right means thinking of all the ways in which AI can help you and creating capabilities at the intersection of business and technology, not hope for a technology vendor to do it for you or people with casual knowledge of the business to be able to pull off a business or technology to be pulled, to be able to pull off good stuff.
Karan Girotra: So that's what I'm seeing with big companies. they're making some mistakes in these initial things. Nothing that can't be fixed and in a way it's good they got started, but there's a long way to kind of get to the full benefits.
Chris Wofford: When we were discussing this, you had mentioned that a lot of responsibility is, heaped on the tech groups within an organization. Maybe there's a deficit of education that is going on across the organization. Maybe the buy in phase, maybe people are skipping the strategic and going right to the tactical phase.
Chris Wofford: What's the problem there?
Karan Girotra: I think it's a little bit of all you said. The way technology works these days is: People put out consumer products and that creates hype. And then people try to just kind of copy them into their business. Oh, there is great Chat GPT. My kids are using it to do homework.
Karan Girotra: Let me make it available to my employees to do emails. Not bad, not incorrect use, but it is a small, small fraction of what you can do. What you've got to think about is, Hmm, can, how can I fix a business process by doing this? How can I change things in the organization to do this? And I think this is all still about big companies.
Karan Girotra: If we think of individuals, if we think of small and business, small and medium sized businesses, I think they're at a different stage. They don't know how to get started. We have some things to say about that, but in principle, I think on big companies, it's a little bit of both.
Karan Girotra: They're starting out in just copying the consumer applications. Businesses are richer. Businesses have more and different impacts of AI. We got to think of that. And then I think you nailed it. We're all in the business of education. I think that's why we're doing these events. That's why with eCornell, we've put out several courses.
Karan Girotra: I think the education around AI, both on the technology side and the business transformation side, if I may say is, even in the leading companies, there's only a very superficial understanding of what the technology can do and what the path to business transformation is. Or all the possibilities of business transformation that exist.
Karan Girotra: So I think education definitely is a missing piece right now. And yeah, and I think we just copied the consumer stuff. So we're thinking in a, in a small way. And this is again for big companies. Individuals and small and medium businesses are in a slightly different place.
Chris Wofford: Tell me about where, small businesses and individuals are.
Chris Wofford: And we can, you know, tell me about those two to start.
Karan Girotra: Yeah, for individuals, I think I have a pretty direct ask for or direct kind of recommendation there. If you haven't done so, number one, whatever professional stage of career you're in, whatever kind of organization you work in, or you're just an individual who doesn't even work at a company or anything like that.
Karan Girotra: This is a powerful technology. You've got to get to know about it. And the best way to get not to know about it is: Just start using it. Not as an old toy, I'll go play with it, because somebody sent me something on WhatsApp about it, or I heard on Facebook how people are using it. Just take a normal task that you have to do, and see if you could, ChatGPT could help you with it.
Karan Girotra: A few months back, I would have said, you gotta buy a subscription. The good news is, ChatGPT is making their Frontier model available mostly to people for free. So try it. And now I also warn, when you try it, it will be a little bit less impressive than all the cool dudes on Twitter who are talking about it.
Karan Girotra: It will be less impressive because those things, people are selecting and showing off cool stuff to kind of generate the hype. But, but it won't be nothing. Play more with it. Give it some time. Some of my colleagues say you need to give it 5-8 hours of tinkering before you kind of get a hang of what it works.
Karan Girotra: Buy a subscription. They're not very expensive. 10, 15, 20 dollar subscriptions, at least for a few months. Try using it. You might be frustrated, but keep at it. That's the first level, I would say, for individuals. If you've gone through that stage of using the ChatGTP, ChatGPT conversational interface, there is other cool stuff which is available with a subscription, not an expensive subscription, but the same subscription that you can tinker with.
Karan Girotra: So most of these companies will have co pilot uh, playgrounds, so to say, places where you can go and design your own co pilots. It's a web interface for, for ChatGPT. I think it goes at platform.openai.com. You go in there and I think you can, you can, you can just start creating your very specific version of ChatGPT, which has access to your documents, your life stories, what you need.
Karan Girotra: So you can tinker quite a bit with the stuff that is very, very easily accessible. If there is one thing I could say about AI technologies, compared to all previous technologies, they're much easier to use and much easier to build applications around. I would say with a few hours of googling and copying and pasting code, you can build an AI powered application.
Karan Girotra: These companies are making these things very easy to use. What is hard with AI is to figure out where to use it, where not to use it. But in per se, it's easy to use. So as an individual, you've got to start using it in all its iterations. The conversation interface, the more sophisticated design interfaces, and then I think you'll get at least the technology awareness to start thinking about where to use it.
Karan Girotra: And then I this is a plug, but I will say this because I think it, I believe this very much. I think in the end, enroll in a course. Enroll in a course. We've put out several courses around both the technical sides of building AI and around business transformation sides, generating AI for productivity, AI and business transformation, we're doing these things to make make the knowledge accessible because of our frustration with the somewhat superficial things that are going on in both small and big businesses.
Karan Girotra: Get going, enroll in a course, you'll learn more and do not think the technology use is a barrier. I think that's not a barrier here. This is easy to use technology. If you understand what it does, you combine it with with some training on how you can use it. I think you can make the magic happen and at least get started on that on that journey.
Karan Girotra: This is, I think, what my recommendation for individuals would be.
Chris Wofford: Yeah, two things. I, absolutely appreciate the idea that it's easy to use. You have to figure out where to apply it to, right? So there's two ways to approach this. This has been my experience and maybe you see this in your, in your work as Cornell faculty, a teacher of this.
Chris Wofford: it's actually really important to just start using the tool and then the application or the solution or the problem that you're trying to solve will then kind of reveal itself. Through the use. And I think the education thing is really important. We'll drop a couple certificate programs. I mean, Karan, you're working on courses with, for Cornell University, live, you know, synchronous and asynchronous courses.
Chris Wofford: All the time in order to keep up with this. So we'll drop some URLs for that. But I really appreciate the answer. So also, let's talk about small businesses, too. I think you hit it a little bit. But for a lot of people, they imagine vast barriers of entry, right? This is going to be really costly.
Chris Wofford: We've got to do things in our business that we've never done before. And this is 10, 20, 100 people, right? How can they apply AI?
Karan Girotra: So let me address costly head on. It costs 30, 20, roughly that amount to get an Open AI GPT subscription. Pro subscription.
Karan Girotra: With that and with these kind of slightly hidden and less obvious interfaces beyond the conversational interface, I think you can just with that, with the playground, you can build your own co pilot that is pretty powerful. Already. You can do quite a bit. So I think the cost is not prohibitive in this stuff.
Karan Girotra: I always say this even with big companies, this is not something where you need armies of engineers to code it. This is stuff which is because the model of all of this AI is to the hard work is what OpenAI is doing. And they're trying to make the interface to connect to that that engine that can do the hard work very easily accessible through the web, through different interfaces.
Karan Girotra: And AI can help you code even some small things, so I don't think you need a large IT team or anything like that to kind of implement many of these ideas. I have seen, I have worked with companies, 15 people company, they have one tech guy or tech person, and that tech person is able to build in pretty useful co pilots for them.
Karan Girotra: So I think this cost technology. Oh, I need IT People. I need fancy people to build this is just not true for this technology. Get started, and you I think you will find that it's not hard to build and the cost, at least to put a put out some proof of concepts are modest of the order of hundreds of dollars, not even thousands of dollars to just try something.
Karan Girotra: Tinker with something. Next thing I will say is the same thing I was saying for big businesses and big businesses are missing the boat in terms of thinking of all the ways in which AI can do. They're thinking of mostly AI consumer like assistance. Small businesses also can kind of, can think beyond that.
Karan Girotra: We don't need to think of AI can do many things for us. At an individual productivity level, yes, it can help you do the tasks you do. Either by delegation or helping you with those tasks. That's just one way. The second way in which AI can help you is, it can make some business processes, work processes more efficient.
Karan Girotra: Think of you're a contractor. Something painful you've got to do. Every year you've got to put a bid out, you've got to file some code, some application permits for getting some code, you've got to make five calls, you've got to send a report on this, this, blah, blah, blah. There is some stuff you've got to do.
Karan Girotra: Pick a business process and see, how can AI improve that business process? In our courses for processes, we talk of several kind of templates in which they can help. The simple ones are automated. Oh, this writing this email, we can use ChatGP to write this email, but that's not it. There is many more things you can do.
Karan Girotra: You can use some AI tools to gather information about projects you've worked on. Perhaps they can summarize the comments you're getting on the, on the projects you're doing, you could have AI listen into your conversations and very quickly give you kind of, okay, these were the specs that this, this particular person client wanted to be built in their house.
Karan Girotra: Just take those notes for you. You can use AI, not just to automate, but to gather more information. If you look into small business processes, half the time is spent in connecting information from one technology system to another system. Or they send me something on email, now I gotta cut paste into the drawing software, now I'm gonna do this, this.
Karan Girotra: You know what? AI is a great connector of different interfaces. Of different software packages. Use it like that. AI is a way to make your make yourself more flexible. You can kind of put put out rather than putting out one design or one pitch. You could get help use AI to help you produce 44 alternate designs for a particular client.
Karan Girotra: You can kind of generate things faster and the lower cost to do that. So I think those are specifics. And we talk about those specifics in many of our courses, but what I would encourage people to do is don't just take the lens of an individual assistant for you, take the lens of business processes, and how those business processes could use this technology to automate, to gather more information, to have better integration in different aspects of the process, to make the process more flexible, more modular, and with that, just try some experiments.
Karan Girotra: Try some experiments and you will see, you will start seeing some, places where, it can affect you. It can improve your productivity quite a bit.
Chris Wofford: Professor Girotra, so tell me if I'm hearing things right, let me tell you what I'm thinking here. Maybe what you're saying, I'm reading between the lines.
Chris Wofford: You tell me if this is off base. I'm hearing that we really shouldn't focus on, let's say ROI. In the way that we might traditionally do so, I think it's big business, right? Large businesses, rather. You had suggested earlier that maybe we're looking at the wrong KPIs or whatever. And this goes all the way down to the small, medium, individual level. Should we be focusing on ROI? Is that a healthy way to look at this? Because what I'm hearing is have new ways and new efficiencies yet to discover, and you have ways to transform your business that you don't even know you know about yet.
Karan Girotra: I would amend that a little bit.
Karan Girotra: I wouldn't say don't look at it. That's a dangerous road to go down because the minute people stop looking at ROI, they, then we have fantasy projects, which, which just don't work sometimes, or you have silly stuff going on. So I wouldn't say abandon ROI, but I think, what I'm saying is what you've done so far is very little and maybe you haven't done it right.
Karan Girotra: So don't, kind of say it's done with these first experiments. So do it right and then look at ROI. And doing it right means not just thinking of AI as an assistant, but thinking of AI as all the other things you can do, as you, as you kind of pointed these things out. So a lot, lot of things AI can do.
Karan Girotra: We are not even in the first innings of the transformation. And at this point, the ROI you're seeing may not be reflective of all that can be done with that. I think that's a fairer statement than saying abandon ROI. It's just that what you've seen so far is just not representative of what, what all might happen for two reasons, because there's a lot more to be done.
Karan Girotra: And second I think a problem for big businesses is not enough technology and business kind of understanding, small businesses on the good side. Everybody in a small business kind of needs to understand the business. So I don't think you have these isolated technology groups or things like that in the small business.
Karan Girotra: So I think they have less of the intersection problem, but I think everyone needs to know this is a, there's many things AI can do, and we've just scratched the surface right now, so the ROI right now is just is not necessarily predictive of all the things you could do.
Chris Wofford: Fine answer to a rather coarse question, so thank you for that.
Chris Wofford: Professor Karan Girotra, thank you so much for joining me again today.
Karan Girotra: Thanks guys. Looking forward to next month.
Chris Wofford: Me too. Can't wait. Thanks viewers. See you next time.
Chris Wofford: Thanks for listening to Cornell keynotes. Check out the episode notes for information on eCornell's AI courses and certificate programs. Thanks again, friends, and subscribe to stay in touch.