Higher education is sleepwalking into the AI era. Students are already using AI, employers are already expecting AI fluency, and companies are already reorganizing work around it. But many colleges are still treating AI like a cheating problem instead of a civilization-level shift in how people learn, work, create, and compete.
Tim’s story makes the deeper point: the future belongs to people who can see a problem, teach themselves what they need, and use AI to build solutions fast. He went from kinesiology and martial arts to coding, entrepreneurship, AI tools, White House conversations, and leading systemwide AI strategy that supports 97,000 students because he followed need, purpose, and persistence.
The most important lesson is not “everyone should use ChatGPT.” It is that education has to stop assigning work students can fake and start designing work students actually have a reason to care about. If students cannot answer, “Why should I do this? What’s in it for me? How does this prepare me for my future?” then AI will become the easiest outsourcing tool in history.
The transcript is also a warning to faculty: AI probably will not replace good teachers first. It will expose mediocre teaching first. Courses that ignore AI may lose students to courses that help them build chatbots, solve real problems, use multiple models, fact-check outputs, and develop future-ready judgment.
The biggest institutional failure may be this: K-12 students are beginning to receive AI literacy, employers are demanding AI literacy, but colleges may become the place where AI literacy goes to die. That gap is not just embarrassing. It is dangerous for students.
So the real takeaway is blunt: AI is not coming for education someday. It is already here, and the schools that keep pretending otherwise are not protecting students. They are preparing them for a world that no longer exists.
Intro & Meeting Tim at Lone Star College
Stefan: Hey, Tim. Welcome. It was nice to meet you. It’s now been almost a month, or slightly longer than a month, since we met down in Houston at Lone Star. I know it was at the online college, but across all the colleges, Lone Star has 90,000 students. So that’s pretty impressive.
I also know you’ve been teaching in higher ed, I think, since 1993. I know your background is kinesiology, which is not computer science or instructional technology, but you became one of the most visible faculty members in higher education, not just across your university but nationwide. You were even invited to the White House, and you had two commercial tools that were shipped before ChatGPT even existed.
So, how did you get into all this?
Tim: I do get asked that question frequently because people say, “Kinesiology and AI? There’s a disconnect there.” I took one computer class in my lifetime, and that was freshman year of high school. I earned an F. That was my only training. But then, in the late 90s, I needed a website for my martial arts gym.
Martial arts has been my lifelong passion, and I attribute many of the things I’ve been able to succeed at to my martial arts training, with all the discipline and the never-quit attitude. Anyway, I needed a website, so I became a beta tester for Microsoft’s first HTML tool. I taught myself HTML and built a website. I wanted automation, so I taught myself programming and added those features.
Then that expanded when I saw some pain points that needed to be filled in the market. So I started writing commercial software and selling software worldwide, basically.
From Kinesiology to Coding: An Unlikely Journey
Tim: About six months before ChatGPT was released, I learned about OpenAI, and my brain blew up. I thought, “This is not going to change just education, but the entire world. I’m going to learn as much as I can about it.” I have fully immersed myself since that day. Every single day, I have never missed studying and learning more about AI.
So I already had my first product, the EIT, that I started developing probably around 2015. That came from a strong need I had as a faculty member.
Traditionally, in kinesiology, when there’s an activity component required, faculty would require students to turn in a spreadsheet or a Word document and write down what they did for their workouts. As I would read these workouts, I would think, “Okay, copy and paste, copy and paste, copy and paste,” and I knew they weren’t doing it. Then, speaking with my colleague at Lone Star College, Ronnie Despeka, he said, “You know what? We should have them wear Fitbits or activity trackers.” So we did that for a semester, but then we had all this data coming in.
It was like, “How in the world do you decipher all these different styles of spreadsheets and data?” That’s how EIT was created. I needed a tool to take all this different activity tracker data and compile it into a grade based on set criteria. It was incredible how it enhanced our kinesiology program.
Then I commercialized it and started selling access to schools around the country. I did integrate AI into it. For example, students have to track nutrition for three days and write an analysis paper. So, into EIT, I built a platform where students enter their papers. It gives them feedback as a tutor would, not rewriting it for them, just guiding them. “Hey, you said two things about this. You should say three.” Then they go and rewrite it.
That was released, like I said, before ChatGPT came out. It was blowing people’s minds, just that one feature. Then I expanded it to my whole platform, Pocket AI, which has that same feature. On the faculty side, they put in their rubric and assignment. On the student side, students get that feedback.
It was creating all types of things, anything a faculty member would need, with very little prompting. They just say grade level, topic, go, because all the prompting was done behind the scenes through the API. So I guess that’s, I don’t know if that’s the short or the long version, but that’s essentially how I got going between kinesiology and AI.
Anand: Tim, what I love about the story you tell about getting involved in all of this is that it was really driven by your interest and your need. You found a problem. You had a problem, and you wanted to find a solution to it.
We’ve talked quite a bit about this, and there are many others talking about it, especially with AI, that we need to teach entrepreneurship, or really problem-solving, to students.
What is it about your background, do you think, that gave you that when the tools weren’t quite as easy to use? I mean, teaching yourself HTML and then some programming to do what you were looking for, that’s a lot more friction than most people would be willing to put up with. What gave you that sense of perseverance?
What really motivated you to work in that way? And the crucial part, then, is how do we take that and model it for our students? I think it’s going to be really important in an AI-inflected world that they have that same mindset.
Tim: I really love that question, and the answer to that question is the basis for almost everything I do. I would say, in the beginning, what really stemmed from that and pushed it forward was, for example, I did an instructor test for Muay Thai, which is a sport in Thailand. It’s like boxing, but you can also hit with the elbows and knees.
Building AI Tools Before ChatGPT
Tim: And kick them anywhere. So I did that test, and I got the crap beat out of me. I ended up with my thigh bone broken open. I grew an extra bone in my thigh muscle as a result of the beating I took. But my goal was that I had to throw 60 kicks and 40 knees each round against two different opponents. They had pads on, but their goal was to knock me out, and I could only hit their pads. So it was very challenging.
I got knocked down a lot of times, and I got back up every time. At a certain point, I knew I wasn’t going to pass. I wasn’t going to get all my requirements in, but I kept getting up. After that test, which is on YouTube if anybody ever wants to see it, people would always ask me, “Why did you get up? You knew you weren’t going to pass. You knew once you got up, you were just going to get beat up again.”
The answer is, “I’m not going to quit. I’m going to finish, and I’m going to go until that bell rings.” That never-quit attitude is something that has stuck with me throughout my life. Like you said, back in the 90s, trying to learn HTML on your own was very difficult. There weren’t online resources for it.
I had basically one of the first web pages. I eventually built the world’s first martial arts discussion forum. I had over a million visitors every day. I built it into this huge thing. But it was hard because I would have to go to the bookstore and stand there in Barnes & Noble reading books, then buy some. That was a hard way to learn. But what really drove me to do it was the need. I saw the purpose.
The purpose was that I needed a website for my martial arts business. I wanted to be one of the first to have one, so I did it. With programming, I had a need. I needed specific things to happen, so I taught myself how to do it. When I took the programming class freshman year of high school, I really did earn an F because I did nothing. It made no sense. How does a zero and a one result in this?
I didn’t care. I had no purpose. I had no interest. They were making me take it. Well, what am I going to do with this? So, zero application, zero effort. When I did have a need, that’s when I put the effort in and made it happen. And for everything in my business, I’ve been like a serial entrepreneur. I’ve had many businesses over the years. I opened my first martial arts gym at the age of 18.
I’ve had many other businesses since then, and I’ve taught myself how to do everything because of the need. I would like to relate that to assignments. When students don’t see the purpose or a reason for doing it, it’s really easy to outsource that to AI, or to your roommate, or to someone else to do that assignment for you. But if you have a purpose and a goal, that’s something you want to do.
The example I like to give is entrepreneurship. If students are in a business class and have a dream of owning a business, and the assignment is to create a business plan for their dream business, do they want to do well at that? Do they want to do it? Are they excited for it? Yes, they are. But if they have no desire for business, it’s, “Hey, ChatGPT, write this business plan,” and they’re done.
So that incentive is a driving force that I think we need to integrate into all curriculum as much as possible, and get faculty thinking along those lines. The way I like to say it is that students need to be able to answer three questions: Number one, why should I do this? Number two, what’s in it for me? Number three, how does it help prepare me for my future?
If they can answer those questions, you’re going to eliminate some cheating right there. Cheaters are going to cheat. We’re always going to have that.
Tim: But the idea is that you give students a really good platform, a really good opportunity to learn what they want to learn and what they see benefit from, and they’re less likely to cheat.
The Power of Purpose-Driven Learning
Stefan: A couple of things stand out from your story. One is that I had one computer science class in high school. I was interested, but I didn’t fail. We just had one. I think it was C. Then I didn’t do anything with it in college.
But when I was getting my education master’s degree, they offered some kind of session in the middle of the day. I remember it was three hours, like “How to Build a Web Page.” So I was like, “Oh, I’m going to go to this.” I learned some basic HTML, and I was like, “Wow.” I built this page.
Then there was a class where the only thing you had to do was a project. So the only thing I did for the whole class was make an HTML web page that had the differences between the U.S. and Iranian education systems, which were pretty big at that time. The professor was so amazed that I could do this that she gave me an A+, even though it only took me about 30 minutes.
But that also got me my first job. I was a debate coach and a computer science teacher, even though I only ever had one computer science class, and I taught web design.
So you never know where things are going to go. But back to the question of motivation. You have a keynote where you ask, “Who uses generative AI seven days a week?” Generally, it’s very few people. Five days a week? Four days? For many, it’s zero, or now maybe a little bit. Frequently, there are a lot of faculty who don’t seem to want to use it or integrate some aspect of it into their courses.
How do you think we could increase that motivation among faculty?
Tim: Yes, very good question. I’m happy to report that when I ask that question, the percentages are increasing. I see it across the nation. But the first public presentation I did on AI was in March of 2023, just a few months after ChatGPT came out. When I asked that question, nobody had used it. I was like, “What?” I was shocked.
In my mind, I was thinking everyone was on this. Everyone was exploring it because it was going to change everything we do in education. So they were on top of it. When nobody raised their hand, I was just in disbelief. Then a month later, three months later, six months later, I was still seeing very little involvement with AI. I was just like, “How is this possible?”
This is going to change everything we do, and you’re not using it? You’re not studying it? To me, something like this is like, how could you not? It’s going to change education more than anything ever has, and yet you’re not using it or even learning about it.
Anyway, I do see those numbers increasing. What I see really helping faculty get on board is, you know, I’ve seen so much resistance over the last few years. People will actually get in my face, angry and cussing. One faculty member said, “Why can’t we use DeepSeek?” Because in Texas, the governor said it’s illegal to use DeepSeek if you’re in a college institution.
I said, “Well, that’s the governor’s prerogative.” He said, “Well, you need to fix that.” And he was cussing, saying, “You need to fix that BS.” I’m like, “Wait a minute. First of all, why are you upset with me? I have nothing to do with it. And number two, why don’t you fix it?” My gosh. But anyway, I just laugh when people get riled up like that because I have nothing to do with it. I’m here to help.
What really does help faculty turn that corner, especially those who are resistant, is when they see how they can use it in their classes to make their classes better, engage their students more, and produce better outcomes.
Martial Arts, Resilience, and the “Never Quit” Mindset
Tim: Once they start to learn and understand the processes they can use and how they can modify assignments, they get excited and get on board. They say, “Oh my gosh, I’m revamping my entire class. I’m integrating AI.” They’re excited. I haven’t seen that excitement for years because they’ve been doing the same thing year after year.
Now, here’s a chance for something new and exciting that will benefit students if integrated correctly.
Anand: The experience you had in March 2023 definitely resonates with me and Stefan. I remember we were working on a book, and we actually published it in March 2023. We had the same perspective. It was like, this is changing everything. We have to get this book out immediately because education is going to be totally different by the fall. Then we got to the fall and realized people still weren’t really using it.
So it really hadn’t changed much. It hasn’t changed that much since then. That’s been eye-opening for us, to realize that we’re early adopters. You’re an early adopter. Maybe there’s a bit of a benefit there because we’re able to see what the technology is, and we have a sense of it.
There’s a little bit of a curse, too, because in some ways that jars us out of everybody else’s reality. I’m wondering, thinking about that, what is it about the way you see AI changing education over the next year or so that other people aren’t getting yet? Maybe the other part of this is that I’m wondering if people will take us seriously, given that we were all that off base in March 2023.
Not that we were wrong. It’s just that our timeline was totally wrong. I’m wondering how you negotiate that, and what you would see as coming up next that is probably going to be disruptive in education, but that most people don’t see yet.
Tim: The things that I see, and you see, and Stephan sees, I think are different from what the reality is going to be. What I think we see happening a year from now is probably going to happen two years from now or three years from now because it’s just going to take that long to get people on board and really start to make those changes. Things move slowly in education, as anyone who’s been involved with education understands.
I thought that by now, several years into this, things would be much different. I know it’s going to happen. It has to. There’s no choice. The people fighting against it are fighting a losing battle because eventually that fight is going to make them extinct.
That leads me to something I often get asked by faculty: “Am I going to lose my job to AI?” The answer I always like to give is, if we think about how many students find your course or decide to take your course, they may go on Rate My Professor and read, “Course A: We did three quizzes, wrote four papers, and the class was good. It was fine.” Then another class says, “We built chatbots. We did engaging things with AI. I could see how I’m being prepared for my future. I loved it. It was exciting. It was new.” What happens next semester when students read those reviews? Where are they going to take a class?
Then what happens to Class A enrollment? That enrollment dwindles. This other enrollment increases. Eventually, they’re not making class loads. So really, they’re phasing themselves out of a job, losing out to faculty who are innovative and integrating AI into their courses. That’s one pathway I see these things going.
Why Students and Faculty Lack Motivation
My big concern is perception: perception from students and perception from industry. Those involved heavily in education understand the importance of education, and even with AI, why interaction with faculty and formal education are still important. But the question from students will increasingly become, “Why should I go to college?”
“Why should I learn how to write a paper when I can write a paper on my phone?” I had a Texas A&M student several years ago pull out his phone and say, “Why should I learn how to write a paper when I can write a paper with one click?”
Tim: That’s the perception of students, and it’s going to increase. Then there’s perception from industry. Why do we value this four-year degree when potentially the whole thing was done by AI? What’s the validity in this degree? Those are the two concerns, and both are perception-based. Maybe not reality-based, but perception-based.
So we need to do a good marketing job and demonstrate through actions to students why we are still valuable. We need to do the same thing for industry and demonstrate why we are still valuable. If we don’t, it’s very easy for industry and students to say, “We don’t need college. We can just ask AI.”
Stefan: You mentioned this before, and you mentioned it again. When you first started using it, you were like, “This is going to change everything.” That’s kind of how I initially got involved. It was early December, and I had just put the article up when I did a presentation on Friday. It was a December 6 article in the New York Post.
I saw an article that said this thing was going to replace Google, and I was like, “Oh, I’d better read this. I use Google all the time.” Then I saw it was ChatGPT. I had noticed language models before, but they were just turning out gibberish. I thought, “Yes, this is going to change everything.”
When I put this article up in the presentation to explain how I got involved, one of the people on the board was a teacher. He said, “Look at the date on that.” He said, “That’s December 6, 2023, when Stephan learned about this.” He said schools didn’t learn about this for like two more years after that.
I thought, well, they kind of did. My perception is that they kind of did, but they just ignored it. But maybe I’m wrong. Maybe they just didn’t see it at all.
A-RAD.
When you talked about helping faculty redesign their courses and redesign assignments to be more interesting, I think one of the things you came up with is called AI-Responsive Assessment Design, or A-RAD. I wonder if you could talk about what an assignment might look like before A-RAD, and then, if a faculty member were to use this with the same learning objective in mind, what that assignment might look like after.
Tim: Sure. I developed that framework in 2023 because very early, and again before ChatGPT came out, I saw the writing on the wall. Very early, I started thinking about how I was going to redo my assignments because if students can just give a prompt and get the entire assignment, I need to do something to keep the assignments legitimate.
I noticed that as I thought about assignments, I was going through the same thought process. Then I took that thought process and developed it into the A-RAD framework.
Essentially, the way that framework works, and the way my thought process worked at that time and still works, is that I look at an assignment and ask myself, “What is the ultimate objective of this assignment?” Once I can answer that, because we have objectives for assignments, but really, truly, a lot of those things are put in there just to check boxes. What is the true objective of the assignment?
I answer that, and then I ask myself, “Could this assignment objective be met while integrating AI into the assignment?” If the answer is yes, great. Let’s redesign it and integrate AI. If the answer is no, figure out why not, because maybe it can be modified. And then, of course, there’s a maybe. I do understand and agree that not every assignment should have AI involvement, but some can very effectively.
So, to answer your question with an example of before and after: a before assignment might be, “Write a paper about the effects of fitness on health.” Something simple. Write a 500-word paper.
Faculty Resistance to AI: What’s Changing
Tim: After A-RAD, and this is actually an assignment I do in my courses, I absolutely love this assignment. I give students the option to choose, and I think that’s really important. I could say, “Do an assignment about cardiovascular fitness.” Okay, they may not care about that. But if a student has diabetes, they may want to know, “How can I improve my health if I have diabetes?” Or, if they’re an athlete, “How do I increase my athletic performance?” Those are things that interest the student and that they want to learn more about. But if I feed the topic to them, they don’t have that care.
So I tell them, pick any topic of interest that has to do with fitness, health, wellness, nutrition, anything. Go out and research it. Find three sources that back up that interest. With those three sources, use that as your knowledge base to create a chatbot. They create a chatbot. Then they have to write the prompt. They have to have a conversation with the chatbot.
The knowledge sources are already in there as part of it. After they have the conversation, they have to fact-check and supply three citations. So now they’re doing research and fact-checking. Then they write an analysis paper about that experience, what they learned from it, and what the takeaways are. Then they turn all of that in.
What I love about it is that it gives them the opportunity to explore something that will be beneficial for them and will help them. They have to research. They have to learn how to create a chatbot. They have to write a prompt. They have to supply knowledge sources. They have to do the chat. They have to do an analysis paper. They can see that it’s very comprehensive.
There are critical thinking skills, analytical skills, and researching. They’re learning about something they’re interested in, and they’re learning how to build a chatbot, which is a really great skill that can help them in many things they do. A chatbot can help them. It gets them started understanding more about AI, which, in many cases, they will have to use in their future jobs.
I loved that assignment so much that I created a chatbot assignment creator, which is just a ChatGPT GPT. All you do is go there, put in the discipline and grade level, and it creates a very nice, comprehensive assignment for faculty. They can then edit it, put their own words into it, add their own flavor, and then share that as an assignment. But I love the assignment. It’s a very nice makeover from the traditional, “Hey, write that paper.”
Why We Overestimated AI Adoption Speed
Anand: One aspect of this that I think is a common thread is the understanding of student motivation and also the way you’re preparing them to develop some AI literacy and fluency with these tools.
This takes me to something you shared with the White House task force: that 97% to 20% gap. I might have the figures wrong, but it was that an overwhelming number of hiring managers are expecting AI literacy, while only something like 20% of students feel like they’re getting it, and that education is providing that opportunity for them so they feel prepared.
When you’re thinking about this, I think you’ve argued that one course isn’t enough. You really have to embed this across the curriculum. This ties into something we were talking about before, about faculty, and making sure there’s motivation for faculty.
But I’m also wondering what you see as avenues for students to develop that fluency so they feel prepared. That sense of ownership over it, I think, is an important component.
Tim: Sadly, I think for the most part, students are learning about AI and becoming AI literate on their own. I think they’re watching TikTok videos, talking to their classmates, experimenting, and figuring things out. I don’t feel like there’s enough official guidance for students in many colleges and school systems. That will change over time. It has to, but it’s coming very slowly.
You mentioned different disciplines. It’s great if you can have a course on AI for students in college. That’s good, but I don’t think it’s great. I think what is great is having maybe a 10-minute snippet that can be offered in every discipline multiple times throughout the semester, because the way AI is used is different in different disciplines.
There are so many purposes, and the limiting factor with AI is really your imagination and creativity. But an accounting class will use AI differently than a kinesiology class.
So if snippets of AI literacy and AI training can be embedded into every discipline where it is applicable, which is many, students benefit. It’s not being shoved down their throats, because if they hate it, or they have opposition to it for environmental concerns, religious concerns, or whatever it may be, it’s not like they’re having to do an entire semester of a full course.
At Lone Star College, I recently met with the first-year experience class, and they interviewed me for a short video that would be shared with all faculty teaching this course to set some expectations and maybe provide some guidance on how they can integrate AI into the first-year experience class. I think that’s a really great place to start, as soon as students get into college.
Here are the basics. We’re also working on a micro-credential that will have AI embedded into it as well. But above and beyond those initiatives, which I think are fantastic, I really want to see those discipline-specific embedded AI literacy snippets. I think that can make a big difference.
Stefan: Yeah. I try to explain, and usually don’t have time to get to this part, that GPT has a couple of meanings. There’s the technical meaning, generative pre-trained transformer, but there’s also general-purpose technology. It literally applies to everything. I think academia’s first experience with it was kind of a negative one, at least from their perspective. It was just, well, it outputs text, and students can write a paper.
But it’s so much more than that and has so many implications beyond that. Like you say, getting into the different courses and subject areas and helping students learn how to use it is really essential.
The Future of Higher Ed and Who Gets Left Behind
Stefan: When I was there, if I remember correctly, you were asked by the chancellor to work on some special projects. They gave you a lot of options, right? They gave you a year to come up with some special projects and work on them. You said, “Well, I want to put together an AI task force for all the campuses.” I should clarify that there are seven or eight live campuses and then the online school.
So you put this all together, again under your own initiative, supported by the chancellor. You also advised the Texas Higher Education Coordinating Board. There are different levels. We talk about faculty learning more. In some instances, they just haven’t had as many opportunities or time to learn. In other instances, there’s some resistance.
That’s at the classroom level. In putting the task force together, obviously supported by the chancellor, there’s support at the higher administrative levels of higher ed, but in other places there’s probably some resistance, or maybe things just move too slowly, if it’s not outright resistance. I wondered if you might reflect on the role you’ve had across the system.
Tim: Sure. About a year and a half ago, the chancellor reached out to me and said, “Hey, Tim, I’d like to present you with an offer. I’ll put your department chair position and faculty position on hold for one year. You come work out of my office.” I said, “Well, what would I be doing?” His answer was, “Identify systemwide issues and solve them. As long as it doesn’t break board policy and it’s not illegal, you can do it.”
I was like, “What? Identify systemwide issues and solve them? What the heck?” He didn’t say anything about AI. That was never part of the conversation.
I said, “May I think about it?” He said yes. So I actually thought about it for well over a month because Lone Star College has 8,000 employees and 97,000 students. To identify systemwide issues and solve them, oh my gosh, that is a huge task. Eventually, I reached back out to him and said, “Yes, I would like to accept the position.” Again, still nothing about AI.
Then, last January, the position started. The first week I sat there, I was like, “What in the world am I supposed to be doing?” I was just so perplexed. But a need I identified was that a system of this magnitude needed a central AI task force. So I thought, “Okay, I will create a systemwide AI task force.” I created a proposal for it, kind of a plan.
I never really had a discussion with the chancellor about my plan. I did have a proposal that he, I don’t know, I hope he read, but I will say he was not opposed to it because I was never shut down or stopped. I went forward very strategically.
The way I did that, even before I was on that special assignment, was that I was already doing things for the college, like offering workshops, doing presentations on AI, and trying to build up the AI champions. I wanted to get them on my side and behind me because I knew that, no matter what, having those AI champions would help spread throughout the system organically.
That was really my first step. After that, once I had my plan, the next thing I did was ask for a meeting with the vice presidents. I went to the vice presidents’ meeting and shared with them. I said, “I am creating an AI task force, and I would kindly ask you to recommend two AI champions from your respective campuses. That is who will comprise the systemwide AI task force.”
Then they asked me a bunch of questions, as I assumed they would, and they liked the answers. They all sent me two recommendations for who should be on the task force. I then reached out to some different administrators. I contacted one of the campus presidents and said, “Would you like to be co-chair of the AI task force?” Immediately, yes, he was on it.
There is another college president who is a member of the task force. I got IT involved. I got a librarian involved. Staff, too. I wanted to get a nice cross-section of people, but only AI champions who are for AI. I didn’t want any of the naysayers on that task force to slow things down because this needed to move fast. That’s the only option.
The idea is that the naysayers can have their chance after we start building momentum, and then I would love to hear what they have to say. But I just didn’t want that negativity in something that had to happen quickly. Once that got going, and it’s still going today, I still chair that task force.
The next part of the plan was that each of the eight campuses needed to have an AI council. So the representatives on the system task force started AI councils at their respective campuses, and they built up groups on their campuses. What happens is we have campus-level meetings. Then the chair from each campus comes back to the systemwide one. They share what’s going on at every campus. That’s how we start every meeting. “Let’s get the campus updates.” They share what they’re doing, and then other campuses can get ideas.
AI-Responsive Assignment Design, or the A-RAD Framework
Tim: They say, “I love that idea. We’re going to do that at our campus.” Then the system team gives feedback to the campuses. It’s really great because we have information flowing from the campuses to the system, and from the system back to the campuses. That structure has worked out really well. I’m very pleased with how it’s all running.
After I went to the VPIs and got their buy-in for this, the next thing I did was go to the faculty senate presidents’ meeting, and I told them this is what’s happening. That strategic order of operations was very important because I never faced any resistance anywhere along that path. Thinking about it, I feel that is remarkable, to not have any resistance with a system that huge.
There certainly is resistance out there, but there was no resistance in that pathway of getting this thing started. I think all colleges need some type of AI council or AI task force. The hard part is getting it started and getting buy-in.
But if you follow that process, if you don’t have one at your campus, I think following that process of first getting the faculty champions involved, starting low, then jumping to the high level, the chancellor or the president, getting their buy-in, and then filling in the middle grounds after that is a good approach. It certainly was successful for me.
Anand: What I love about that is it really resonates with something else I wanted to ask you about in terms of agentic AI, and it gives me some insight into your thought process. It’s really systems thinking. You’re planning out a system.
You’re thinking about different constituencies and different roles they’ll play, not just putting them all together and randomly figuring out how they’re going to contribute. I really like that order of operations. The way you did it maximized the output while encouraging lots of different perspectives.
I’m guessing that a similar perspective drives what you’re doing with agentic AI. I’m really curious about this, and I think Stephan’s probably told you a little bit about it, but a lot of our work has been with AI pluralism, this idea that we need multiple perspectives from AI models, different kinds of AI models. I was wondering if you could speak to that a little bit.
As I understand it, in some of the agentic work that you do, your workflows don’t just have agents that are all the same model. You use different types of models. You’ll use OpenAI’s, or Gemini, or Claude. Why do you do that? What’s the benefit of incorporating them in different ways and then orchestrating that agentic system?
Tim: Yeah. I can’t remember exactly when I set it up, maybe seven or eight months ago, but I created a system where I’m using a free program called Microsoft Visual Studio Code. In Visual Studio Code, there are extensions that you can enable. So I enable Gemini, Codex, which is ChatGPT, and Claude. Now, that’s fine. That doesn’t make it agentic just by doing that yet.
Then I created Markdown files that contain instructions. The instructions basically say, and this is very simplified, of course, “When I say use the multi-agentic workflow, you will consult with ChatGPT, Gemini, and Claude. You will all spawn multiple agents to work on the same problem. Once the solution from each agent is output, Claude is the boss. You are the boss, Claude. You will analyze each of the three outputs and give me the final plan and recommendation, and justify it.”
So once I give it some type of complex prompt, they all go to work. I get a table. Codex says this, and all the solutions it has. Gemini says this. Claude says this. Then there’s the justification from Claude as to why it selected the final plan.
As I read through that, it is just so interesting. The justifications are interesting, and they will pick up on each other’s mistakes and errors. The final output is so strong and so much better than if just one of them worked on it all on its own. Using that system, I’ve been able to become so incredibly productive.
Actually, let me step back. Before I started using this agentic workflow, when we talked about entrepreneurship, I’m still amazed I was able to do this because I could never have done it without AI. Like I said, I opened my first gym at age 18, and I’ve owned many different businesses since then. But about a year ago, maybe, it was a Friday night at 6:00 p.m. I was sitting here and thought, “You know what? I have no plans for tonight. What am I going to do?” So I decided I was going to start a new business.
Between 6:00 p.m. and midnight, six hours, here’s what I was able to do. I created Evolve AI Institute LLC. I did all the LLC paperwork, submitted it, and got my business checking account set up. I built out a complete, very extensive website. I built a detailed 12-month marketing plan. I built out all the ads. I got the business cards ordered. I got the logo designed and finished.
A Better Assignment: From Paper to Chatbot Creation
Tim: There were other things, too, but every single thing you can imagine for a complete LLC business company was done between 6:00 p.m. and midnight. Finished. Ready to go the next morning. To me, that’s just remarkable.
I was teaching an entrepreneurship presentation on how you can use AI in business. I asked, “Who’s created an LLC?” Some hands went up. I said, “How long did it take you from start to finish, just the paperwork, getting through all that?” He said, “Four months.” I said, “Well, guess what? I did it in an hour.” He was like, “What? How’s that possible?” That power is incredible.
In the last few months, I can’t even tell you how many platforms and businesses I’ve built. I did an experiment once I created my agentic workflow. I gave it a prompt before I went to bed. I said, “Do market research to find a gap in businesses that don’t exist but should because they can be monetized. The stipulations are that the platform has to be delivered from a website. It has to essentially be self-sustaining because I don’t have a lot of time. Create a whole new business for me. I’m going to bed, so don’t bother me with questions. Claude, you are in charge of answering.” I clicked submit and went to bed.
I got up the next morning and read the most remarkable plan for a new business that doesn’t exist, but it should. If I had more time, I would be doing this business. That’s my problem: time. But it’s sitting there waiting for me. When I have the time, I’m going to do this business because it’s that good. I read it and thought, “This is remarkable.”
So I was thinking, every night before I go to bed, I could give it the same prompt and wake up every morning to a brand-new business idea. That could never be done in the history of the world until now. To me, that’s exciting. When you harness that power, you can accomplish remarkable things. Then you have faculty saying, “You can’t use that.” It’s like, what? The things I can do with that are life-changing.
We have students who have the potential to run million-dollar companies within a month. It’s just incredible. So we need to help. I do feel it is faculty’s responsibility to help guide students with at least the very basics of AI, because if we don’t, we’re doing them a disservice.
Anand: Well, if you need help with some of those businesses, Tim, send them our way. We are happy to pitch in. That’s exciting. I just love that idea of being that creative and using the tools to say, “Where can I go? How can I identify problems? How can I solve them?” Of course, there’s a monetary benefit, but just thinking about society, and how we could benefit our communities by taking that same approach, is awesome.
Tim, I do feel the need to give you one piece of advice. You know how you say Claude’s in charge and we’re going to do whatever Claude says? You probably shouldn’t tell the faculty that when they give input, you’re just going to turn it all over to Claude and do whatever Claude says, right? Even if it’s a good idea, you might want to hedge on that.
But that does get me to one other question. You gave this TED Talk. I think it was called “AI Will Teach Us All,” right? In the title, who would you say is the “us”? Most faculty are like, well, the “us” is the students. But is it also the faculty, the institutions, society at large? We’re so used to being the ones who deliver the knowledge to the students, but it seems like AI is going to teach more than the students. I don’t mean it is going to teach more than the students and replace us, but it is going to teach everyone.
Tim: Yeah. Everyone has a different lens. If a faculty member were to read that title, they’re going to say, “Oh, it’s going to teach me.” If a business CEO reads it, they have a different perception. But the real answer is everyone. It’s going to teach everyone.
In fact, I was recently reminded of something I said in a presentation back in, I think, March of 2023. A person came up and said, “You said something in your presentation. I wrote it down, put it on a sticky note on my monitor, and I share it with my students every semester.” I said, “Well, what was it?” He said, “You said that AI will expose the mediocrity of society.” I said, “Oh, I don’t even remember. I forgot I even said that.”
But when you look at it that way, AI will expose the mediocrity of society. When AI can do things better than us, it’s going to show us that the thing I’m doing here is pretty mediocre compared to what AI is now, or later will be, capable of. There are brilliant people all over the world, but even so, AI is going to become more brilliant.
I was invited to OpenAI headquarters, and I was very fortunate to have a conversation with the vice president of OpenAI. We were talking about AGI and things that AI can do. He told me a story that I thought was pretty cool. The researchers at OpenAI were working on a research project that they couldn’t figure out. There was a gap in the research that was stopping the project.
He gave their system 24 hours of compute time. When we do a prompt in ChatGPT, a few seconds later we get an answer. Well, this system was working on their problem for 24 hours straight. Throughout that process, it found a study from 1960 and related that 1960 study to their current problem. It solved the gap they were missing.
What was most remarkable, he said, was that their high-level researchers couldn’t find that 1960 study. He said that even if they had found it and read the research study, they would not have been able to make the correlation between that issue and this issue. But AI did, and it allowed them to keep moving forward.
He said, “What is AGI? At some things, we are already there. We have achieved AGI. At other things, we haven’t. It’s going to take more time.” But in either case, it’s progressing very rapidly. Things are changing, and they’re going to change even more as time goes on and these systems get better.
The AI Literacy Gap: 97% vs. 20%
Anand: Tim, we really appreciate you taking the time to share with us and have this conversation, and also for exposing the mediocrity in so much of what’s going on in education. For anybody listening, where can they find out more about your work? Where can they find you online?
Tim: LinkedIn. I’m on LinkedIn. They can also find me at evolveaiinstitute.com. I think those are probably the two best places. My email address is accessible from either place.
Something I did want to share, too, and I had a little bit of a conversation with Stephan about it, is that when I was at the White House AI task force on education meeting, it was led by Melania. It included the secretary of labor, the secretary of education, the secretary of energy, CEOs from Microsoft, OpenAI, IBM, and all the huge corporations. They shared their initiatives for what they’re doing to contribute to AI and education. Everything was K-12.
The Google CEO said, “I’m going to donate this much money for the initiatives for K-12,” and so on. That meeting lasted somewhere around an hour. After the meeting, there was a three-hour reception. At this reception, I made it a point to talk to as many people in the room as I could. “Hey, what do you do? What do you do? What’s your plan?”
I found out very quickly that I was the only educator in the room. It was 100% politicians and CEOs. Once I realized that, when I was able to speak to some White House staff, I said, “You know what? This is really, really concerning to me because nobody knows the inside of a classroom better than teachers and faculty. Certainly, they know it better than politicians and CEOs, because they’re there doing it every day. So I really hope you’ll get more educators involved.”
I was shocked I was the only educator there. I couldn’t even believe it. I went to this White House meeting thinking it would be so cool because I would get to interact with other faculty members, find out what they’re doing, and learn from them.
There was one CEO who taught grade school 15 years ago. There was a White House leader who was involved with a college 10 years ago. That was it. That was the only connection.
The other thing I expressed to the White House staff that I found extremely concerning was that these White House initiatives are K-12, yet we have college students graduating and entering the workforce, and they don’t have these same initiatives. They are so strongly needed because students are going through two-year and four-year colleges being told, “You can’t use AI. You can’t do this.”
Then they get to the workforce, and they’re told, “You have to use it.” The story I love to tell is that the CEO of Shopify, which is a hundred-billion-dollar company, released an internal work memo. He didn’t release it, but it got leaked.
Once it got leaked, he went on X and said, “Hey, since this was leaked, I’ll share it anyway.” The two things that really stuck out to me in this memo were, number one, using AI on the job is an expectation. He went on to say that employees using it are 10 to 100 times more productive than those who are not. Number two, before you hire someone new, you need to justify why AI cannot do the job better.
So you have Shopify and other companies in other industries doing similar things, yet we have students we tell, “You can’t touch AI. It’s banned.” Now, with this White House initiative, we have students starting in kindergarten learning AI literacy, and they go K-12. Then they get to college and they’re told, “You can’t touch it.” They’re like, “What world is this? This is just craziness.”
Then they progress into industry, and they’re told, “You have to use it. You have to use this AI tool embedded into Excel to become more productive.” If they say, “I don’t know how to do that,” how do they even get through the interview? How do they maintain the job if they’re not on the same level as their competitors?
So I just think it’s so important that faculty really take a close look at how they can contribute to the success of their students in the future they’re entering. If they don’t, it’s going to be really hard on our students, and that’s the number one priority: our students.
Why One AI Course Isn’t Enough
Anand: You put that so well, Tim. That’s great. Thanks for leading that charge. You’re right. We need to get more faculty at the table, and we need to make sure our colleagues are coming forward to figure out what’s best for the students. That’s really what it’s about. All right. Thank you so much, Tim. This was great. Thank you. I really appreciate it.
Tim: I appreciate the invitation. And let me just add, Stephan, I’ve been following your work since day one, when I saw you on LinkedIn. I’ve always had high respect for your writings and the things you’re doing. So being here is a great honor.
Anand: I appreciate it. It was great to meet you and see everything you’re doing as well. Thanks.