Six months is not a long time. But in the AI and education conversation, six months turned out to be long enough to change the “AI vibes” for educators at the AIxEd conference.
Hamit Hamutcu has had the unusual vantage point of convening two conferences in that window, the first in November 2025, the second in May 2026, both under the banner of AIxEd, the event and community platform he founded to bring together K-12 teachers, university leaders, ed tech founders, policymakers, NGO leaders, and state officials in the same space at the same time.
The first conference felt like a room full of smart, capable people who were still trying to understand what was even happening. The second felt like a room full of people who had accepted that the answer was incomplete and decided to get to work anyway.
“Day and night,” Hamutcu says, then catches himself. “Maybe a little too much. But it really, really was a significantly more advanced nature of conversation.”
Jason Gorman, who attended both, puts it differently: in November, there was disorientation in the room. In May, there was traction. People still did not have the full picture. But they had jobs to do, they had experiments to run, and they had enough of a foothold to take the next step.
That shift, Hamutcu believes, is worth paying attention to.
Listen to the podcast on your favorite directory:
What the Conversations Are Not Yet About
One of the most revealing observations in the episode is the one Hamutcu offers about where the energy in the AI and education conversation is concentrated, and where it is notably absent.
There is a great deal of conversation about how AI is changing the way teaching and learning happen: tutoring tools, assessment workflows, feedback loops, classroom dynamics. That conversation is maturing quickly. Educators are sharing examples. Best practices are emerging. The confidence in those rooms is rising.
The conversation that is lagging is about what to teach.
“There’s a lot of good conversation about how the teacher operates, how the classroom works, how assessments happen,” Hamutcu says. “But maybe not so much on what we teach and what students should be learning.”
The reason is not hard to understand. The uncertainty about what the job market will look like in three to five years is immense. If you do not know which roles will exist, which skills will be valued, or what the entry conditions for a career will be, the question of what to put in a curriculum becomes almost impossibly complex. And so, understandably, many educators are not looking directly at that question yet.
Hamutcu’s view is that they will need to. The feedback loop between the labor market and the curriculum has always been slow, but the AI acceleration is making slowness more costly. Institutions that start adjusting now, even imperfectly, will find the next adjustment easier than institutions that wait until the signal is undeniable.
Context Is Not Optional
When Gorman offers Hamutcu a hypothetical, a magic wand to make any curriculum compulsory across education, the answer is not a subject or a tool. It is an approach.
AI fluency taught in isolation, as a standalone course, a separate module, a skills add-on that sits beside the real curriculum, does not produce the outcome institutions are hoping for. What produces that outcome is AI taught in context: in the marketing class, where it changes how marketing problems are solved; in the engineering program, where it changes how structures are modeled; in the data science degree, where it changes how analysis is conducted from the first week.
“The curriculum, the syllabus, needs to integrate that really intimately,” Hamutcu says. “Not superficially, but intimately.”
The reason this matters is not just pedagogical. It is practical. Students who learn to use a tool in the context of a real problem develop judgment about that tool. They learn when it is useful and when it is not, where it produces reliable output and where it requires skepticism. Students who learn the tool in the abstract develop facility but not judgment. And judgment, as Hamutcu notes, is exactly the skill that does not get replaced.
What Data Science Figured Out the Hard Way
Hamutcu is not new to the problem of a profession that cannot agree on what it is. Roughly fifteen years ago, he co-founded an initiative to bring taxonomic standards to the data science field. He conducted surveys of nearly a thousand professionals, interviewed dozens of hiring managers and chief data officers, and produced a framework that looked like a periodic table: elements organized into groupings, some about tools, some about foundational methods, some about the cognitive process of structuring and solving a problem.
His punchline is both encouraging and cautionary. The first category, tools and technology, has changed almost beyond recognition. The second category, understanding how models work, is partially relevant depending on whether you are building technology or using it. The third category, the ability to structure a problem, identify appropriate methods, and evaluate the results: that one is timeless.
“Those are obviously timeless,” he says. “That’s maybe what corresponds to a durable skill in a more technical setting.”
The lesson for learning professionals is not that taxonomy is futile. Hamutcu believes standardization creates a target state, provides training frameworks, and enables meaningful assessment. But it has to be built with the humility that the parts most worth protecting are the ones that would survive the technology being removed entirely. If a skill only matters because a particular tool exists, it is a tool skill. If it would still matter if the tool disappeared, it is something more durable.
Stop Reading About AI Every Day
One of the most practically useful moments in the episode is Hamutcu’s pushback on the relentless information pace that surrounds AI adoption.
He has unsubscribed from newsletters. He does not track every benchmark release or model update. And he does not think most professionals should either.
“I don’t think the average professional needs to read about AI every day,” he says. “Things might be changing day-to-day. But if you’re three weeks late in using the latest release of whatever tool you’re using, that’s fine. It’s not the end of the world.”
His recommendation is a monthly wavelength. Subscribe to two or three newsletters: something macro, something slightly technical, something specific to your own role or industry. Read them when you can. Miss a week without panic. And then, critically, when something catches your attention because it connects to a real problem you are trying to solve, stop reading and start doing.
That doing is the part most people skip. And it is the only part that actually builds capability.
“Unless that rolling up of sleeves happens,” he says, “people cannot understand the power and also the limitations or the risks of using AI.”
The metaphor he reaches for at the end of the conversation is a train. The train is moving. You do not have to be in the front car. But you do need to be on it. Because the longer you wait to board, the harder catching up becomes.
What Comes Next
Hamutcu’s prediction for the next two to three years is deliberately calm. Workflow integrations will mature. The questions people are asking now about how AI fits into existing systems will settle. Many of those systems will simply have AI embedded in them, unremarked, the way electricity is embedded in everything that plugs in.
What he is less certain about, and more interested in, is the individual initiative story. The most compelling things he has seen and heard across two conferences have come not from institutional decisions but from individual educators who decided to try something and shared what they learned.
Institutions need to provide safe environments for that experimentation, responsible governance, clarity about what is permitted, protection for both students and educators. But the energy for change, he believes, is coming from the bottom up. And the question worth asking is not whether AI will transform education. It is whether the people inside education are on the train when it does.
Listen to the full conversation with Hamit Hamutcu on The LX Ops Show or wherever you listen to podcasts.
Podcast Transcript– Jason Gorman with Hamit Hamutcu
Jason Gorman (00:01)
Welcome to the Learning Experience Ops Show, where we talk about making learning better that’s future ready. I’m Jason Gorman, the founder and CEO of Jackrabbit Learning Experience. The whole purpose of this show is to talk with people who come at our work from different angles with different backgrounds, different vantage points, because from where I sit, that’s how you actually get a feel for where we are right now and where we’re headed.
And a lot of my guests have spent their careers inside education, really deeply immersed in it. But my guest today is a little different. He comes out of data science and AI. And that background gives him an unusual window into this very moment that we’re all living through. But there’s something else. He spends his time convening people from every corner of education. K-12 teachers, university leaders, ed tech founders, policymakers, all in the same room.
So he isn’t just watching the conversation about AI and education. He’s standing in the middle, listening to all of it at the same time. And that gives him an especially interesting read on the zeitgeist. And that’s exactly what I want to talk about today. My guest today is Hamit Hamutcu. Hamet is the founder of AIx Ed, an event and community platform that brings together all the folks that I mentioned a moment ago.
To make sense of what AI means for education. He came to that work after a long career in data and AI rather than education itself, more than 20 years across industry and consulting, including building and leading a global analytics practice, and most recently serving as senior advisor and director at the Institute for Experiential AI at Northeastern University, where he led programs to help organizations build their data and AI skills. He also co edits the Industry Column at the Harvard.
Data science review. Hamit, welcome to the show.
Hamit (01:53)
Yeah,
thanks so much. Thanks for having me, Jason.
Jason Gorman (01:55)
Yeah, thanks for doing it. Well, let’s get warmed up with a couple lightning round questions. These are just just quick to get us going, get us sort of stretched out a little bit. All right. So first question what is the 30 second elevator pitch about what you do?
Hamit (02:11)
Yeah, I mean I I think you made a great intro for that answer. what I’ve been trying to do for the last several years is trying to find and create value right at the intersection of academia and industry.
I think there needs to be a lot more. And that definition of what’s that, you know, a lot more. The answers that I’ve been working on are for people to talk, to collaborate, to jointly work on projects. And I’m trying to facilitate that. AIX is certainly a great example of it. I’m really happy that you know we can convene this kind of unique group of people. My work at Harvard Data Science to do is exactly that. Can we you know create more opportunities for industry and academia to collaborate from a content perspective?
So there’s benefit to both sides. And anything I do and I’ve done for the last several years, and even looking into the next one or two at least, is trying to stay right at that kind of sweet spot of academia and industry.
Jason Gorman (03:09)
And I’m curious just to get like a a little a little glimpse into your origin story here. W what was there an event or a moment that really inspired you to do what you’re doing now?
Hamit (03:20)
I mean I wish I had you know a a great story there, but it’s really, really simple in my case. my roommate in graduate school, he signed me up for an interview at FedEx. I didn’t even know. so I ended up interviewing, I got the job. that got me into data analytics, you know, by by you know, completely by chance.
so it’s really a luck, I guess, and and and circumstance that put me on this trajectory. But I’m extremely happy to be on that you know on that path. I haven’t looked back, I haven’t looked for an alternative. I’ve been just coming at it from different angles over the last 25 years or so.
Jason Gorman (03:56)
In your work in general, when you’re really locked in, what are the things that you’re doing when you’re just completely focused in a flow state?
Hamit (04:04)
Yeah. I think what gets me in the zone is when I’m piecing things together,
Jason Gorman (04:10)
Yeah.
Hamit (04:10)
I mean not really thinking or brainstorming, that’s not what I mean, but you know, connecting the conversations I’ve had, the bookmarks, you know, the notes.
and bringing it together to create some structure out of it. It could be you know it could be a new product idea it could be a new you know event idea whatever whatever it is but can you know can I kind of connect a lot of these dots that I’ve been you know collecting in my brain bring them together and then create something out of it that is greater than you know the parts right so that’s that’s I think is what focuses me most.
Jason Gorman (04:47)
One question in two parts. What are you most anxious about for the future and what are you most hopeful about?
Hamit (04:53)
I don’t have a lot of anxieties. I don’t know, kind of a naive answer, but I mean I’m sure there’s gonna be a lot of ups and downs, but I think I my my working assumption is that things will work out you know in the end. so really not much. I mean I I just try to think of different scenarios and then you know prepare myself and people I work with you know accordingly.
you know, in terms of hope, you know, I of course share the concerns you know that’s everyone or a lot of people have been
really worried about you know about ai and its impact all of that but i I mean I feel like there’s still a lot we can do as as humans and on on how that story unfolds. so as I said you know I I think short term is definitely going to be maybe a bit blurry and and and a bit nerve-wracking but I think in the end we are going to find great ways of using this new powerful technology to our benefits.
Jason Gorman (05:49)
That’s great. And this this could be about work. This could be about something completely outside of work. But what is a skill you’re trying to learn right now?
Hamit (06:00)
I’m not sure about a I’m not sure about a skill. but as you said, I’ve I’m fairly new to education, right? I mean, I talk to these people who’ve been for decades you know in the different domains, they know this inside out, they’ve been in all kinds of different organizations and subsectors, etc. So I am trying to just absorb as much as I can, especially about the educational infrastructure.
you know how the different parts work together. I mean K through twelve and high red, you know, complete different beasts. being at Northeastern really gave me a lot of insights, but I know that I’m still scratching the surface. So I try to make every conversation, including this, a learning opportunity. so I hope they will add up to some
you know, something greater and and put me in a position where I’m able to do more than I am able to do now, say a year or two from now.
Jason Gorman (06:52)
Well, that’s the perfect segue into my next question. I so as you’re having all of these conversations, and I should say I’ve been to both. You’ve had you’ve had two conferences. I’ve been to two conferences. They are both incredible. I learned so much. I met a lot of people, made a lot of good connections, really felt each time like I left with a better sense of the the kind of pulse of the of the kind of where people are and
What is happening and what people are thinking about and what’s important and what to pay attention to? It’s very really useful. So thank you for those. As you reflect
Hamit (07:26)
Yes, sure. Thank you.
Jason Gorman (07:28)
back on the conversations that happened at the conferences, but also around your work with the conferences, what feels important to you right now about the conversations we’re having about AI and education?
Hamit (07:43)
I
mean one thing that was very clear to me, you know, in the the two conferences you mentioned, they were basically six months apart, right? So the first one happened November twenty twenty five, the second May twenty twenty six. I was genuinely surprised how
much it was different, the nature of the conversations. I mean you certainly expect progress, right? I mean the tech AI technology is moving fast, so you you do expect, I guess, some some fast progress on the side of you know implementers and educators etc but it it was very very clear there was a lot more maturity in the way people were approaching you know the topic, a lot more confidence, many more examples.
on how people have
H have used it. I mean, still a lot of questions, certainly. But like I said, I think six months ago we wouldn’t have had half the stories or or you know some of the best practices, etc., that we heard you know last month, you know, in May. so I think the level, the maturity of the conversations what is what would say, the nature of the conversations is is was I mean, I you know, day and night maybe a little too.
Much, but it really, really was a significantly more advanced nature of conversation.
Jason Gorman (08:55)
I have to say, I’m gonna co-sign that. I felt like in the November event, there was still every every presenter obviously knew what they were talking about, was a hundred percent you know, had answers, had interesting points of view. But the the vibe in the room, the conversations I had with other people was still a sense of disorientation.
That we were in a disoriented moment, people still trying to understand what is even happening and what exactly are we supposed to do. And I a month ago, the sense was maybe we still don’t know, but we have jobs to do. And we now have a sense of how to get some traction and we are at least doing specific things to get there. So I really felt that, even more than hearing it. It just felt very different.
as a group. So I
Hamit (09:42)
Yeah, no
I fully agree. Yeah, fully agree.
Jason Gorman (09:45)
Yeah.
so I mentioned before all the people that you bring together just very, very diverse, both the speakers and the attendees. So I met teachers and university provosts and ed tech founders. there was a state senator there, there was a secretary of education there. There was a head of a a large AI, mature AI organization called C Sale. I think is that how you pronounce it? Yeah. Just in a
Hamit (10:13)
Yep, C Sha from
MIT. Yep. Yeah. Yeah.
Jason Gorman (10:14)
Couple days. Lots of amazing people in
in the same room over two-day period. So as you think about all of these great folks that were in the room, what would you want to communicate to somebody who’s listening to this podcast about what you see? What would you want learning professionals to understand if they weren’t able to get there? What would be a takeaway or a couple takeaways?
That represents the kind
Hamit (10:41)
Yeah, I you
Jason Gorman (10:42)
of diversity of thought.
Hamit (10:44)
know I mean first of all I’m really happy you feel that way because that’s exactly what we were aiming to do, right? So we bring people from different parts of education. And the reason why I think that’s really important. And of course if someone was there in person like you were, I’m I’m glad to hear that it was clear.
but for people who are not there, I s would definitely want to highlight that there’s a lot to learn from each other, despite the fact that it might look like our environments, our challenges, our you know, classrooms, our environments might be quite different from each other. I mean, one thing I learned over a couple of decades of management consulting is you know, you would walk into say a bank and they would say, we’re very like we’re very different, we’re very unique, like we’re nothing like the other banks.
Jason Gorman (11:26)
Yeah.
Hamit (11:27)
Or
you you you you or you talk to a bank and they say, I mean we mentioned maybe a work from say a telecommunications company and they say our industry is extremely different. And of course they are. I mean banking is different from telecommunications, but there’s so much more common to it, right? Especially
Jason Gorman (11:39)
Yeah. Yeah.
Hamit (11:42)
when you’re dealing with technology, its impact, the barriers, human nature, you know, change management, you know, adoption. There are just a lot of things that are common.
And I think in education, we should not be creating even more silos, like, this is
Jason Gorman (11:57)
Mm.
Hamit (11:57)
the K through 12 room, and this is that’s where we discuss K through 12 challenges. of course, there are challenges that are unique to K through 12, but I think a K through 12 professional being in the same room with a higher ed person, an ed tech person, a policymaker, an NGO leader. I think there’s a lot of benefit in in making those conversations. So, what I would say is, of course, hopefully you know people will show.
up you know in the next event but if they’re not able to I would definitely consciously you know find ways to connect with people not only in my immediate environment and learn from them which is certainly useful but also you know kind of expand that network of people that you talk to that you learn from because I think there’s a tremendous amount of value in that.
Jason Gorman (12:41)
Yeah, there’s a real danger. I so I spent the first ten years of my career in higher education and then I went into consulting.
Similar to yours. If we just take higher education in particular, it can become an echo chamber. And my experience at the AIX ed conference is that it wasn’t. We were all just there hearing things that we might not have heard otherwise. We’d never have the opportunity to hear otherwise. It was great to have that the potential for so many rich conversations. But even so, so I am curious.
Are there places across education where you think the energy in certain spots is higher and the gains are a little bit better? Whereas in other corners of the education landscape, maybe not seeing the gains with AI use or maturity, and maybe the energy is a little lower around its adoption.
Hamit (13:36)
Yeah.
I mean I so I I’d like to see, you know, I’d love to see how you think about this because you were there, but you know, there’s obviously a lot more action and application in thinking about or doing things differently on how we teach or how students learn, but maybe not so much on what we teach and what they should be learning, right?
Jason Gorman (13:57)
Mm. Mm.
Hamit (13:59)
so you know, say tutoring or assessments or you know, so a lot of conversation about you know how the classroom works, how you know the teacher kind of operates, interacts with students, gives feedback, assesses, etc. So there’s a lot of I think good conversation there and a lot of good examples.
But maybe because of the uncertainty about what’s happening in the job markets, you know, are jobs disappearing, which ones, how are they changing, etc.? I don’t know if people are just blocking out kind of the noise there or they’re too worried and they’re, you know, kind of, I don’t know, not maybe looking inwards and not wanting to think about that. which I understand, by the way. I mean, if there’s so much uncertainty, how can you react to it? it might take a little time maybe for that feedback loop to to work and for the job market.
to settle, you know, etc. But I think there needs to be more conversation about also what to teach and from the student perspective, what to learn. I mean, there’s obviously you know people talk about durable skills and you know things that will help students and and and you know people who are just entering the workforce ride any expected or unexpected turns in the you know in in how jobs you know play out but I feel like that you know there’s probably
Jason Gorman (15:19)
Interesting. Is
If you if you had a you had a magic wand or a genie or something, and you could
you could make I don’t know, s a certain kind of curriculum compulsatory, right? And in K twelve or higher ed or even at the workforce level. What would be the content you would want? What would you want people to be learning about right now?
Hamit (15:47)
I mean I think anything that helps students use skills in a context, you know, AI related or you know technology related in general, I think those are those are really, really important because it gives them a chance to kind of take the first step in what will probably become their future and their career flow, right? Because if they’re able to take that first step, then they’ll be better equipped to take the second and the third.
But if they’re learning skills either in isolation, like you know, and and you know, not really in an applied manner or within context.
Jason Gorman (16:21)
Mm.
Hamit (16:22)
So if I’m studying, say marketing, yes, I mean, you know, AI fluency, you know, the universities are investing a lot, you know, into that. But that being used in a marketing class, in the way I solve marketing problems or business problems or design marketing actions, that’s really, really critical. So the
the curriculum, the syllabus needs to integrate that really intimately, not superficially, but intimately, which takes time. so again, I’m not saying that’s you know the sh no this you know.
Why hasn’t this happened? I mean, there hasn’t been probably enough time. But I think that’s
Jason Gorman (16:55)
Yeah. Yeah.
Hamit (16:56)
what we should be striving towards because but and that will change, right? I mean, we can’t be using the same type of stuff for several years because that will change. But again, for from a faculty perspective, if they’re able to take a step now towards that, taking that second step when things shift again will be much easier if they wait much longer now, right?
Jason Gorman (17:18)
‘Cause they could be laying the tracks for the principles, which may not change that much over time, even as the technology does change.
Hamit (17:26)
Yes, absolute, absolutely.
Jason Gorman (17:30)
In my research for a conversation, I found out something interesting about you. So 20 years ago, it looked like you were kind of looking out across the data science world and saw that the same title might have meant 50 different jobs and that nobody could really agree on what the work even was. my understanding is that you co-founded an initiative to bring standards to the data science field. And I have to tell you.
This sounds exactly like where the learning where learning professionals are right now. And this is not certainly not an AI thing. This is this was true well, well before AI in our case. And I would say it’s even more true now that AI is confused what we do. And maybe it has also in data science. But I’m curious when you think about maybe you could talk a little bit about that work and what you were trying, what you saw and what you were trying to do there. And I’m curious if you see any parallels with.
how the edu how educators or learning professionals n perhaps need some need some standardization or need need some clearer way to think about our work so we can professionalize faster.
Hamit (18:42)
Yeah. it was a long time ago. Probably not good quite twenty years, but it was a long time ago. I mean the first I think the kind of the mainstream popularization of the of the data science professional or the title data scientist was a our business review article. Tom Davenport, who’s a friend and a colleague, was one of the authors. So they said it’s data science is the sexiest job of the twenty-first century or something like that.
so of course when you say that, then everybody wants, you know, or not everybody, but a lot of people want to become data scientists. So the way I saw it at the time was you know, the only requirement for someone to be a data scientist was to change their title to a data scientist only thing, right? Because who is to say you’re nuts? so that’s kind of prompted that entire research study. I mean, we had you know surveys, about a thousand people.
you know I’ve personally conducted dozens of interviews with hiring managers and data science managers, chief data officers, you know, etc., to create a detailed taxonomy of what the skills should be at different levels for different functions, data scientist, data engineer, data analyst, what makes them different.
and all of that. So we don’t find ourselves in a situation where in the same company, this is an actual example, where two people are calling themselves data scientists, one of them has like an undergraduate degree in business and some marketing experience. The other one has like a chemical engineering degree and a PhD in something, right?
So they could be great both great resources for the company, but they probably are not doing the exact same thing. So we should not be called. So that was you know, that was a study that actually, you know, that you know, that’s was kind of my first
Real experience of working right at that intersection of academia and industry because we had
Jason Gorman (20:28)
Mm.
Hamit (20:29)
to be very rigorous in the way we were approaching it. But it was an industry challenge we were trying to solve. And you know, that got published at Harvard Data Science Review. That was my also original connection to the publication, you know, etc. So I think taxonomies, I mean, are important, otherwise I wouldn’t have done it. But we’re now at a stage where I think even the taxonomy that I worked so hard and diligently.
on
is mostly irrelevant or you cannot really apply it because the job itself is has changed dramatically right so you know what i of sort or certainly i’m no expert of you know of your profession and the the the community
you know, that you’re that you’re part of the learning you know professionals. But it any kind of standardization, especially for skills that may that are maybe more durable, certainly helps because it gives people kind of a target state. it provides training trainers with you know more robust frameworks that they can work with. it also provides for assessments either for jobs or for
education, you know, etc. So I’m all for it, but I think we also should recognize the fact that we are at a stage where things are moving really, really fast. so we should be prepared to throw out a taxonomy out the window if necessary.
Jason Gorman (21:46)
So what does I mean, it could be data science, it could be learning professions, it could be anything right now, because everything is changing quickly for most people. How should we think about what to, as my wife likes to say, set a coffee cup down on it? Like, like you don’t wanna you don’t wanna put your drink down on something that’s gonna move, right? So what is the thing that’s not is there something when we’re thinking about the taxonomy of our work where we can say,
Here are here are the things that are really worth investing in over the long term. Then there’s another category of things that are not as durable but are still necessary, that we are just going to need constant PD. How did you how did you suss those things out in your work? And how what would be your advice to people as we’re thinking about progressing through our careers?
Hamit (22:36)
Yeah.
I mean in data science you know we
you know the the framework that we came up with was with was basically look like a periodic table, right? So we had elements, right? And each element corresponded to a particular you know skill. And we had groupings of elements, kind of like in a real you know periodic table, right? and some of them were about tools, right? So we had Python, you know R Excel, so different different tools to process to analyze you know data. you know we had
Some core skills like math and statistics, right? Probability, linear algebra, so a lot of different types of really fundamental skills that underlie any of these you know professions. we also had you know the ability to define a problem, to break it down into pieces, and then to you know to build some kind of code using whatever.
tool that you that you pick. So the first category is, you know, has changed dramatically, right? The tools and the technology, et cetera. Some of the understanding of how a model works, et cetera, I think it’s still relevant, but probably not so much unless you really are a person developing those technologies and not just a mere user of them.
But that third example I gave, which is the way you you structure a problem and then you find the appropriate methods and tools and and and you can you know evaluate the the results that you’re coming back, etc. I those are those are obviously timeless. So that’s I guess in a more technical setting, that’s that’s maybe what corresponds to a durable skill beyond you know the the the general broader durable skills like communication and you know
Right. So I don’t know what exactly that corresponds to on the on the learning side, but there’s gotta be some level of technical skills that are still mostly resilient to the way AI maybe is being you know used more and more.
Right. So that’s I mean that’s one thing. And like I said, I be hard for me to comment on those. What I will say about AI related skills though. I mean I’ll
Jason Gorman (24:36)
Yeah.
Hamit (24:36)
go back to that to the context, right? So the example I gave about the marketing class and and and all of that. because there’s the other thing I get asked lot is how do I keep up with what’s happening in AI?
That in and itself has become a job. I mean, I
Jason Gorman (24:47)
Yeah. Yeah.
Hamit (24:49)
work in AI and I find it difficult to keep up with what’s happening, right?
Jason Gorman (24:52)
Yes. Yeah.
Hamit (24:54)
so so you the I I I don’t think you know the average professional needs to read about AI every day, right? Because things might be changing on a day-to-day basis. But if you’re I don’t know, three weeks late in using the latest release of whatever tool that you’re using, that’s fine. It’s not the you know, it’s not the end the world.
I think that’s that’s probably you know trying to keep up with daily creates a lot of anxiety. I honestly unsubscribe from a lot of newsletters just because of you know of that you know overload. But what matters is how what AI does in your own context. So I would spend a lot more of my time in using whatever tools are are are mostly most commonly used.
among my colleagues or what some newsletter in my about my profession is recommending. And then spend time to actually use it in a workflow that I do. Right. And I know in the in the s the the the community, the SAC community that you’re running, you’re very also encouraged encouraging about experimentation and sharing those experiments. I think that’s what really matters. So I would spend more time there in you know kind of in that thinking and
And doing space rather than reading and synthesizing what I’m reading stay. Because the wavelength I think that most everybody needs to worry about is not a daily or even weekly wavelength. I think it’s a monthly wavelength. And as long as we’re every month we’re you know we’re doing things that are a bit maybe different, we tried something, it stuck, we tried something, it didn’t, and so we moved on. I think that’s what we need to
Worry about if anything.
Jason Gorman (26:33)
It’s very helpful. One of the things that I have started to think about a lot lately that that it sounds like what you’re saying that this might be a way for us to think about it too. Well, most of the other tools that we use, you know, you named the tools that are relevant to data science, those are tools, many of them, that were built specifically for people who do that kind of work or something very similar. When you show up to an LLM, it is a blank piece of paper, it is a Word document.
That has zero content in it. It can be anything. And I think that there is, because of where we are, we’re not in a place where we have a we don’t have very many robust purpose-built tools for what we do. And I think this is true for lots of other careers. Or if they exist, they don’t have a lot of market penetration. People don’t have access to them. They haven’t gotten used to them. And I do.
I do
I think this feeds into this sense of things moving so fast is because people feel as though they have to keep up with what is AI writ large, because I need to understand it writ large to understand its application to my work, as opposed to just saying, this is this application, then I’m going to use at work. It’s going to be the same application probably for a while. I really need to understand it really well, but I do not need to understand a lot of bigger picture things about AI.
But I wonder if we are still in a place where it’s hard it’s hard to it’s hard to just filter down to that industry specific stuff right now because there’s just so so much uncertainty. The tools are still early stages, I think. You can tell me if that’s fair, but they feel early stages to me still.
Hamit (28:27)
Yeah,
yeah, I agree. And and and I think there’s I mean one of the things that’s I’ve been thinking about, particularly with my colleagues at Harvard Data Science Review, is you know, what what’s a publications role?
This right,
so how can we you know what I kept saying over the last you know half hour or so is you know context and you know using it within your own environment and all of that. But if it’s if your if the content that is being thrown at you, at you know media, you know, media, your newsletters, the Substack and the YouTube and videos and you know the conferences, etc., they’re they’re mostly about you know, kind of here’s what this does, or this is you know.
new and look off capable, you know, etc. Then it’s really up to an individual to you know absorb all of it and then filter out things that are not relevant to their own context, and then
know, you know, whatever is left, work with that. I it’s a lot to ask you know from a from a person.
Jason Gorman (29:25)
It is. I agree.
Hamit (29:26)
so I think you know we all should and that’s certainly one of the things I I’m trying to think hard about is
How do we make that process easier for people? So it’s not just an
Jason Gorman (29:35)
Mm, mm.
Hamit (29:36)
article, another article or another video that we share with people, but it’s really, you know, when they have a question, I’m at work, I have a question and I try and I’m trying to understand how I can use AI for a particular problem, or I have a meeting and I’d like to you know contribute, but I but I don’t know exactly some of the things, etc. Then I you know, someone should be there, like not if not not everyone has a personal consultant sitting across from
their desk who can you know with infinite wisdom and ai seems that way but it’s not because you know it’s it also has a lot of a lot of other things so can we kind of you know narrow it down in a way that makes it very relevant very direct for who you are and what you’re trying to do
Jason Gorman (30:18)
That’s interesting. So are you saying that the review is thinking about filling its role in that way? About being a mechanism or a a tool or a resource that people could use to help with that kind of filtration, simplification, application work?
Hamit (30:40)
Yeah, I mean it’s a it’s a big task. so we’ve certainly been
Jason Gorman (30:42)
It is, sure.
Hamit (30:43)
we’ve certainly been thinking about it, right? Because if you’re an academic journal, there’s a traditional, you know, model that you go with which brings all the credibility with it, of course. It’s
Jason Gorman (30:50)
Right. Right.
Hamit (30:52)
peer reviews and there’s it’s very rigorous, you know.
very selective and all of that, which has its function, but if we wanted to branch out from that, what does that branching out look like to bring all of this rigor and expertise closer to a broader you know group?
and I’m sure it’s not only us thinking about it, but I think it would be it would be great if maybe even within verticals people try to work on that kind of thing. I mean, from being now you know immersed in the AI and education communities is certainly something that I’m you know thinking about because I know for a fact, I’ve heard this over and over again. I mean, if you’re I mean you you you you said you met some teachers, right? So you know firsthand
You might have some fam in your family and you know or your colleagues. No one has the time to do all of this, in addition to a job that’s already extremely busy and taxing and tiring. So I think whatever we can do to make that process, that cognitive process of understanding, synthesizing, and then applying it in context, I think that would go a long way.
Jason Gorman (32:01)
Hmm. Can’t help but agree. okay, everybody wants an AI prediction. So I’m gonna ask it. I appreciate hearing about it from your particular vantage point. what is your prediction for what happens in AI and education over the next few years?
Hamit (32:17)
I
mean I think a lot of the productivity kind of workflow related applications will be will be quite mature.
Jason Gorman (32:25)
Mm.
Hamit (32:26)
some of the kinks will you know will it will hopefully be worked out. the way it’s you know kind of integrates, you know, are you throwing out an entire class of applications and bringing in AI
fueled ones or are you accelerating things by AI? So all of those types of questions that people are asking now, I think most of them will settle down. so you know many of the systems you know will be embedded with you know AI capabilities, people will have reworked their you know kind of main workflows and all that so it’ll be a lot more it’ll probably look and feel a lot more like business as usual rather than
Jason Gorman (33:01)
Mm-hmm.
Hamit (33:02)
you know, the new technology, what do we do with it? Which is kind of what’s what the feeling is now.
I don’t think it’s gonna take too long. So you said two, three years. You know, I think that’s that’s a reasonable time frame. I’m not sure if my earlier comment about what we are teaching and how st or and what students are learning. I’m not sure if that’s gonna be if that’s gonna be all in the clear. I I I think you’ll feel calmer also in that front. I mean, I’m not I don’t think jobs will be disrupted in such a major way over the next couple of years that.
universities will be irrelevant. I mean certainly I certainly I don’t subscribe to that you know extreme you know extreme scenario but if in two to three years we’ll you know we have revamped a lot of the curriculum and individual you know syllabi with you know with with with with with our courses across the world I’m not just saying AI courses or AI fluency or I’m saying all of the degree programs and and whatever whatever it is
that you do. You know, AI is infused, it’s integral. you know, it’s again, it feels more business as usual. I mean, that might be happening in especially more applied areas, like if you’re if you’re studying architecture or if you’re studying engineering, if you’re you know, certainly if you’re studying data science, you know, all of those probably are gonna happen a bit quicker.
because there’s more incentive you know to do that. Maybe other professions or other degree programs might take a bit longer. But I’m hoping in two to three years you know that contextual learning of AI
is gonna be a lot more common than it is now. It’s not a, you know, so I didn’t predict a doom or or or or or or or major explosion of AI is gonna take over everything or whatever. So it I I it’s it’s a more realistic take probably on what’s gonna happen. and probably a bit easier with the time frame that you gave. I mean if you said five to ten years I would I pr my my short answer would have been probably I I have no idea. I don’t know.
Jason Gorman (34:59)
Yeah.
Completely fair. I appreciate that what you’re describing sounds calmer, more settled,
That we start to think of AI as a utility that runs a lot of other things in our lives and not a thing. Like I don’t we don’t walk around thinking about electricity until we don’t have it. And then it is the most important thing. And then it right. I I it sounds like there is a from what you’re describing that AI will become kind of part of the fabric. We still have to pay attention to it.
still is as important, but it may not be this kind of big dominating topic that it is right now in quite the same way.
Hamit (35:49)
And I’m especially optimistic about what individuals are able to do. I mean, that I don’t know if you would agree with that, but a lot of the things that I think we enjoyed hearing actually came from individuals taking initiative to do things. I think organizational catch-up still needs to happen. Like how do we govern this? What are the ethics? I mean, certainly there’s a policy as aspect, although that also, you know, that is, you know, there’s obviously a lot more discussion on that now, which is good. I discussion needs to happen.
but what we know what is okay versus not okay, you know, to do. So I think so I don’t wanna you know you know, I’m not saying I’m you know I I’m not ignorant of all of those really important discussions that need to happen at the organizational level. But I think that individual initiative is just gonna drive a lot of positive change. And organizational just needs to play, you know, make sure that’s all of that is done in a
in a responsible way. I mean we are talking about children and young adults. It is an extremely, extremely sensitive part of our society and you know our future and and so so you know it’s it’s it’s really critical that we get you know things right. but like I said I think that you know there’s a lot of bottom up. There’s definitely thinking and discussion you know up top at either at the policy level or governance you know you know discussions around responsible AI and and ethics of you know data and the privacy issue
that comment that etc. So I’m hoping those will converge so that as people are you know doing things, they’re also provided with kind of safe environments in which to do those types of experimentation.
Jason Gorman (37:23)
You’ve already touched on a few things, just two questions left. my second to last question here, we’ve already kind of touched on a few of these things. I’m gonna ask it anyway because there could be another dimension that you’d want to add. if somebody who’s listening to this conversation is overwhelmed by the nonstop deluge of
technical advances and political decrees about which models we can have access to and AI economics news and general hype and doom cycles. What what is your advice to them right now? How do they stay grounded and ready for what might happen, let’s just say, in their career?
Hamit (38:03)
I mean first of all if someone’s not working in AI, like that’s not their field of work or or profession, then I don’t think this should be a like a daily part of my routine.
I think maybe it’s weekly. I might subscribe to a couple of newsletters, maybe something more macro, you know, something that’s maybe a bit more technical, and then something that’s related to my specific role or industry. maybe I read those on a weekly basis. If I miss a week, that’s fine, I don’t panic. right. and that probably gives me the base knowledge about what’s happening you know in AI.
Kind of like what you would do to keep track of your you know immediate environment, your town, like the community that you’re part of, you know, etc. and then, and I and I said this you know before, but I really do think it’s it’s critical, whatever it is that you’re learning, if something piques your interest, like you read the story, you said, okay, this could apply to my you know classroom, or I have the exact same challenge, then you know you have to roll up your sleeves. I mean, unless
that rolling up happens, I don’t think you know people can understand the power and also the limitations or the risks of you know using AI. so that initial phase of you know
the prompting and how to say it and and to to make you know to get ai do what you want want it to do, etc. I think you know that obviously is becoming a lot more taken for granted now. A lot of people feel comfortable you know with that. but the next step is is is a bit harder. And I’m not, you know, I when I say harder, I’m you it’s not like technically harder because we’re still talking about a fairly easy task for even non-technical people. I’m not saying
Go do vite coding and then build your own application. That’s not you know that’s
Jason Gorman (39:52)
Yeah.
Hamit (39:52)
not the level. Although if you’re feeling like that, absolutely do it. I mean, because it’ll show you a lot of things
Jason Gorman (39:55)
Yeah. Yeah.
Hamit (39:58)
that you wouldn’t have otherwise you know be exposed to. But find that that’s you know next level of of of you know tinkering with AI is absolutely what needs to happen. Because as I said, if you don’t do it now and next month and three months from now and six months from now, and you know, and the train is moving, it’s just becoming harder and harder.
harder to catch up with it. You know, as long as you’re on the train. I mean, because the train might slow slow down, it might speed up, maybe you you you you know you find yourself in a in kind of a a a back car and and and and and that that’s fine. I mean you can always you know you can always catch up but you know people need to you know be doing things right now if they’re not that would be my real real
Otherwise, what happens today with this tool or that tool or what this new release does, what the benchmarks say about image generation or what you know, critical you know, problem solving or whatever, I mean, those are s much less significant.
Jason Gorman (40:55)
Yeah, that’s great. Okay, last question. I always give a question from a person that I’ve speaking with recently. And the one today comes from Brenda Boyd. So her question is, how do you find how do you define quality in your work?
Hamit (41:14)
I
mean I I think once when you do something, of course, especially if something that you put a lot of effort and time into, you would like to think that it’s you know not only is the right thing to do, but it’s also done in the right way, which probably is what we could say is you know quality, you know, quality is but I think the real test is how people perceive it.
right? so if if a product is being adopted by its intended audience, if a conference is well attended by people that it’s supposed to you know attract, so whatever it is, I mean if if you’re finding
kind of the the reaction you think you deserve, then probably you can feel more confident that you’ve done a you know a quality job. If it’s not there, it might still be a you know a a a good product or or offering or service or whatever you know whatever it might be but it probably is missing a few things that’s that that that
basically prevent people from just taking taking that leap. So I’d rather have other people assess what what that is because I know having worked in data science I also know quite a bit about the biases that’s that humans have.
Jason Gorman (42:28)
So many. I really appreciate this conversation. It’s been such a pleasure having you on, Hamit Thank you so much for joining the show.
Hamit (42:36)
it’s been my pleasure. Thanks so much, Jason.