“Shit in – Shit out. Without Precise Input, ChatGPT is Useless to Students,” Says Tomáš Hülle (ECCEDU)

April 4, 2023

What does ChatGPT need to be an excellent assistant? Precise instructions and flexibility in subsequent development. ‘If we don’t teach students how to correctly define their requirements and precisely specify their input, ChatGPT will be of no use to them during their studies,’ says Tomáš Hülle, director and founder of the European Centre for Career Education (ECCEDU).

• What role will universities play in education, why do they often conflict with IT companies in practice, and how will ChatGPT affect this?
• How can ChatGPT assist disadvantaged students in the future, and how will it significantly speed up and facilitate the education process?
• How did American universities handle GPT-4, and how is it that only 14% of them can work with artificial intelligence?

Listen to the podcast (CZ)

Read the podcast as an interview

Ivana Karhanová: ChatGPT is now in its fourth iteration. Recently, a Russian student defended a thesis written by AI in just 23 hours. What does ChatGPT signify? Is it a doom for education, or will it make all tasks effortless? Should its use in studies be penalized, and if so, how? Today’s guest is Tomáš Hülle, the director and founder of the European Centre for Career Education (ECCEDU). Good day.

Tomáš Hülle: Good day.

Ivana Karhanová: Will we have dumb students educated by artificial intelligence?

Tomáš Hülle: Hardly. Or it will depend on how you define a dumb student. However, the educational system’s fear of cultivating increasingly stupid students has existed for centuries. It doesn’t matter whether we’re talking about the invention of the printing press, calculators, pencils, or anything else. The concerns are always there, and I don’t feel we are getting dumber.

Ivana Karhanová: Isn’t it the opposite? Don’t we have increasingly smarter and more educated students who can cover a broader range of information?

Tomáš Hülle: That’s exactly what I meant. Yes, exactly.

Ivana Karhanová: How do you personally view the use of GPT or any other AI tool in studies?

Tomáš Hülle: Since its original release last November, we’ve been using GPT, especially due to our marketing department’s influence. It has significantly facilitated our work. It feels like we are four or five times more efficient.

Many tasks that would have been incredibly complex to obtain and text are now rapidly achievable thanks to this tool. Then, we can further refine it to make sense, perfectly capture the vision, and match our strategy. So, we perceive GPT as a tremendous acceleration, efficiency, and focus on what matters.

Ivana Karhanová: Can you give a specific example of what you requested?

Tomáš Hülle: We were working on a prestigious proposal for the European Commission, where you might be texturing thirty pages. You have fifty to sixty questions, each needing two to three thousand characters. Often, you’re rewriting similar text versions.

Nothing is simpler than giving ChatGPT a basic idea or product vision. Essentially, you explain it as you would to an employee, what you aim to achieve in the project. Then, you input each question and see what the tool creates for you. It’s like assigning it to an intern or another employee. You take the answer, and if it makes some sense, you refine it. If it makes no sense, you at least know how not to write it.

Ivana Karhanová: You mentioned something important. Essentially, you need to describe what you want. That’s probably the key to future use, isn’t it?

Tomáš Hülle: Definitely. This is the role of every director, every company founder, every CEO. Without a vision, without a product, or without the ability to clearly explain it, then IT follies, for example, in development, are created.

I can well imagine IT companies’ concerns when dealing with assignments from large clients. If the client cannot describe what they want, then you will repeatedly rework solutions simply because it wasn’t sufficiently explained. This isn’t a novelty; the recipient of the message is different in this case.

Ivana Karhanová: Because in my LinkedIn or YouTube bubble, of course, examples of what ChatGPT can generate multiply. How many posts for Instagram, Facebook, how many product descriptions. Yet, the principle of garbage in – garbage out still applies. If I give it a lousy task like: write me ten headlines, then those headlines will be about nothing.

Tomáš Hülle: I agree. If we want to educate leaders, and they can’t understand where they are heading or their vision, what they want to achieve, whether founding their companies or managing others, then exactly that principle of garbage in – garbage out applies. That means if they don’t know where they are heading, they can have ChatGPT as much as they want, but they will never get any sensible answer.

And so, education’s fundamental role is to teach students to communicate precisely what’s in their minds. If they can do that, then this tool is absolutely fantastic because it will do the routine work for them, the work that has no added value and is not even enjoyable.

Ivana Karhanová: But if they are capable of giving precise instructions to any system, it means they understand the issue.

Tomáš Hülle: That might be said. But if you have precisely defined instructions, then the question is whether in the case of new products, you won’t find that this precision is detrimental to the development itself. You don’t always realize the end state. To me, it’s about a combination of precise instructions and a certain degree of flexibility in subsequent development.

Ivana Karhanová: Looking at the history of innovations in teaching, in the 1990s, the source of information was the library. Then came the internet, and in between, there were photocopiers. So you could at least photocopy the book, take the pages home. Then students began copying texts from the internet and assembling them, so tools were created to detect if you copied. And now we have artificial intelligence. How do you see its use in education now?

Tomáš Hülle: That is a question that nobody can answer right now. Just yesterday, I was looking at a survey across American universities. How many of them actually have a strategy concerning GPT, not just in chats but in artificial intelligence in general. Currently, only 14 percent know how to handle artificial intelligence within their institution.

Ivana Karhanová: Some even prohibit it, if I’m not mistaken.

Tomáš Hülle: Especially in New York and Seattle. But that’s similar to when the church tried to fight against the printing press six hundred years ago. They can try, they can slow down the implementation, but in the end, the tool cannot technically be banned. Currently, all that schools are doing is blocking access on school computers or school Wi-Fi.

But I remember, I think it was students from Princeton, where GPT Zero emerged as a project. They quickly devised several easy ways to circumvent the bans, from VPNs to other tools. That’s one line.

You also mentioned software capable of capturing work, as Masaryk University tried to do about fifteen years ago with a system called Egg to Egg. The question is whether it’s technically possible at all.

Although I’m not technically minded, when I discussed it with Tomáš Mikolov, who is considered the biggest expert in artificial intelligence, not only in the Czech Republic but globally, he told me he couldn’t even imagine such a tool technologically.

Ivana Karhanová: You just need to set that no more than x percent of the text can match, right?

Tomáš Hülle: I can’t answer. Probably.

Ivana Karhanová: When we talked before filming, you mentioned that artificial intelligence, or specifically the ChatGPT tool, could help underprivileged children. What should we imagine under that?

Tomáš Hülle: I imagine it similarly to how introducing laptops or computers to teaching helped people who couldn’t afford to pay for elite schools, individual tutoring, or simply had individual access from early childhood. It could be simplified in the United States, where there is an elite segment of society expected to succeed no matter what.

When you give more power to those who don’t have these conditions at the start, say by giving them computers, later the internet, and access to libraries, they gain much greater access to information, their own development, and can achieve what they otherwise wouldn’t.

With GPT-4… If the tool is not monetized or is charged at a level that, for example, the state could support, then students will have access to all know-how. You talked about libraries that were accessible only if you were part of the church or some privileged layer. This is different. Suddenly, through GPT-4, education can be obtained by everyone. Now it’s just about being able to use this tool correctly.

Ivana Karhanová: And do you think these children would be disadvantaged if they didn’t have access to these tools?

Tomáš Hülle: Absolutely. Consider how, as I mentioned at the beginning, we were able to write proposals for the European Commission four to five times faster. If a student has to spend four to five times more time obtaining some information, they are inherently disadvantaged.

Ivana Karhanová: You collaborate with universities, offering students help with the transition to practice. What is the current generation of students struggling with the most?

Tomáš Hülle: It depends a lot on the field. To oversimplify, it’s access to experience. Globally, we work with about one hundred and fifty universities, and what universities have always provided was know-how or knowledge, often drawn from books, the written text. And in elementary and high schools, you learn when an event occurred, how many rivers belong to a certain basin, or simply where a village is located or how many inhabitants it has.

Over time, of course, knowledge becomes more and more sophisticated, but what universities have never been able to give students, and it still holds, is experience. Now, it doesn’t matter if we take students of architecture, business, law, or anything else. But we actually try to complement this aspect with universities. That means our mission is to involve companies in education, so students can draw experience directly from practice, from companies, from people who really do these things.

And here we return to GPT-4 again. That’s something GPT-4 will never give you. It will never implement 25 marketing strategies for you. It can describe them to you. But the implementation is something you’ll have to do yourself.

That’s actually the part of education where we try to be active, where we are currently in contact with Harvard, Yale, we have the most successful global program at UCLA, all top ten universities in the world. We try to work on their students gaining experience from practice.

Ivana Karhanová: On the other hand, the fact that students lack experience is actually a generic assumption of who is a student.

Tomáš Hülle: Yes, but at the same time, without that experience, they are not able to make the transition to practice successfully enough. If you are part of the young generation today, you have to move incredibly fast to succeed. The competition, brutal, is global. You gradually need to have two or three languages, have XY internships, have an idea about several different sectors. Therefore, the assumption of experience shifts. And it shifts more and more into the younger years.

I know many students, some of whom I also mentor, who have been doing internships in IT companies since they were fifteen or sixteen. They get these internships because they just storm into LinkedIn, write to the CTO of some company, asking if they could join, if they could help him two days a week. And these students, who gain such experiences at fifteen, are far ahead of those who precisely think that way: We are students; we don’t have to yet.

Ivana Karhanová: And when we talk about IT, Adastra is a data company. How do you see the shift in IT education? Because it develops incredibly fast.

Tomáš Hülle: It does. It’s paradoxical. The question is what the current trend of layoffs in the sector will do. IT was the first sector to massively realize the lack of talent and employees capable of working, whether in coding or in the data field or anything else. And it had to approach universities and find a way to collaborate with them to draw employees or colleagues as early as possible. Other sectors can perceive this as good practice, from the perspective of IT.

The question is how to resolve the conflict that currently exists between universities and IT companies. IT companies often go to universities and tell students, and they mean well, that they don’t really need school and that they will teach them everything in practice. This then causes a clash between schools and companies. The ideal role of a university or its goal is for the student to stay there the entire time and complete school optimally.

And if an IT company teaches the student that they really don’t need school, then the role of the university must somehow be redefined. And we circle back to GPT-4, which puts tremendous pressure on universities to find their role. Because nobody says that for a university to fulfill its goal, the student must stay there until the end of their studies.

Ivana Karhanová: I discussed this with a few people, and their opinion was that the school probably won’t teach you the things you need, but it teaches you to manage stress and gives you a general overview. Do you see anything else from the perspective of the university, what schools should fulfill, especially in IT? Where knowledge changes so fast that by the time you graduate, what you learned as a freshman may already be different.

Tomáš Hülle: The first thing you mentioned was stress, and the second was a general overview, right?

Ivana Karhanová: Stress and a general overview.

Tomáš Hülle: It’s interesting because I paradoxically feel that I wouldn’t expect either of these two values from a university. It seems to me that when a person is in practice in an IT company and responsible for a project, the level of stress in relation to colleagues who depend on it is many times greater than the stress of having a good grade or being fully prepared for an exam.

The general overview nowadays seems to me that students often obtain it everywhere possible, just not at the university. The university is very focused on a specific field. But it’s also true that across the field, students generally gain a broad overview. I would agree with that.

For me personally, the role of a university is in two or three key dimensions. The first dimension is the development of a person’s personality on an ethical level. Simply to make a responsible and complex person out of a human being. The second significant role, which worked amazingly for me at Masaryk University, was inspiration. I met several people there who greatly boosted me in terms of being passionate about something, if I may say so. Radim Polčák, Professor Hajn, they were just amazing people who started me on the next 15 years. That means some inspiration.

And third, which doesn’t quite work in Czech education but is very common abroad, is basically a network of interesting people with whom one can be in contact and with whom one will work in the future. That is the role of Harvard, Stanford, and other elite schools in the West, for example. It cannot be said that every personality will necessarily be an ideal mature personality after graduation. But it can be assumed that many of these students will succeed professionally. And so, in America, one of the key roles is the Alumni Network. In the Czech Republic, there is minimal work with graduates.

Ivana Karhanová: When we touched on what students face when transitioning to practice, it is often said that Generation Z, which is now entering or already in practice, doesn’t want to work much. The view of the generation even older than me is that they are actually quite lazy. How would you see them with your experience?

Tomáš Hülle: I’m lucky in that I can lecture as part of the You Can Do Business project. I have several other high school projects where I’m not only in contact with universities but also meet a lot of high school students. And it seems to me to be complete nonsense. Sorry to put it that way.

Just from the fact that I remember the same argument 10 years back. The same argument was made by the generation even before. It seems to me that this is something we will say again in fifty years, that the upcoming generation is simply lazy, doesn’t understand what’s important, can’t work hard, and so on. For instance, I have one student, and he is absolutely incredible. In one week, he can code an application, plays the violin, does two sports. Tonight, we have an event with Sotheby’s where I invited him to an art exhibition, so he just comes and networks like a 35-year-old professional. And regarding the notion that he couldn’t finish things… If you saw this student, how he develops the application. He can say, look, I don’t know this, so I’ll outsource it. He asks a friend and indeed uses this word. And it’s not because he is lazy, but because it’s more efficient. The way the current generation can use many tools simultaneously to be more efficient, to meet enormous demands, is incredible.

You’d be surprised at what I can imagine where there could be change. As they are subject to high demands and have to constantly learn more and more, young people are less stable in employment. They don’t last 5 or 10 years in a company unless you manage to transform their role in the company so they are constantly learning, constantly advancing. For this generation, it’s typical to constantly feel like they are missing the train. So they always try to catch up, work hard on themselves. But of course, this is very generalizing. You will definitely find lazy members of the young emerging generation. But I don’t think it’s a significant problem.

Ivana Karhanová: Just as you will find lazy members of the older generations. You’ve led me to another thing I wanted to ask you about. What, from your perspective, should companies offer graduates to be passionate about working for the same company, ideally, for several years?

Tomáš Hülle: I think many companies forget to ask at the beginning why the student is joining them. Or if they ask, it’s someone from HR. But to ask the employee every once in a while, hey are you satisfied? Are you developing sufficiently? Is there something we could do differently? Then you can work very well with that student or that representative of the young generation.

When you know their expectations. It doesn’t necessarily mean changing the world. Maybe they just want to progress in their career and earn more money. Then you can beautifully set their progression across the company. But there’s nothing worse than not asking and then being surprised. In two years, the student will want to leave, and it will already be in their mind, just waiting for an opportunity to jump.

And when you have LinkedIn, you have a million headhunters hunting students even through Tinder, then especially in competitive sectors, you won’t keep that student. We did some research among fifteen hundred students from about ten universities in the Czech Republic, and they all talk about impact, what they leave behind, as a fundamental value for the incoming generation. But equally percentage-wise, we found that salary is important to them. That means, I wouldn’t underestimate such classic factors. I’d take it as students wanting basically everything and companies need to adapt to that. And if they can, the student or representative of the young generation is truly excited and can function much more efficiently.

Ivana Karhanová: We started our discussion with artificial intelligence and the ChatGPT tool in relation to what you would improve in the education system.

Tomáš Hülle: Well, that’s an excellent question. How to answer briefly? If I could pick just one thing, then it would be the need to work more on teaching staff. On people who themselves want to develop to further develop students. For them to realize it’s not prescribed that a student must tick off knowing all the protozoa, all the flora. For these people to be able to work with a student similarly to how mentors or coaches do.

And how to achieve that? I don’t have a recipe for how to do it exactly. For example, when we educate students through companies, we look for people in those companies who have that spark in their eye. We are interested in whether they can explain their field to students and are eager to do so. With us, it’s not about paying them XY money for lecturing. Of course, we have to give them some reward for going. But we mainly need people who can deliver the message correctly. Then it’s actually irrelevant whether we talk about GPT_4 or the impact of social networks or fake news or simply about critical thinking.

The things that need to be learned will always change. I read at Berkeley, where they called partner companies they work with as part of the policy hub. Among others, there was a company that asked in this group when the university would include GPT-4 in the curriculum. And the university responded, of course, as a conservative environment would. We know about it, we are working on it. Simplified.

But the question from the technology company was not correct either. Because by the time these students get into practice, another five years will have passed. By the time they work as team leaders, it will be another ten. By then, we will technologically operate on something completely different. So even that reaction from the industry, which will say that we now need to train blacksmiths or now we need to train plumbers, is nonsense. By the time these students arrive in practice, the market situation will be completely different.

Ivana Karhanová: Says Tomáš Žíla, director and founder of the European Centre for Career Education. Thank you for coming to the studio, and hear you next time.

Tomáš Hülle: Thank you very much.