Introduction

From AI tools to the future of education

we can play a little bit more with Gamma and also with Claude, as I promised, is to think about what this means for the future of education. This would probably look familiar as a platform because it's Gamma itself. So that's where I made my cards, how they would call them.

And I mean, there is, you might be left wondering whether we are gonna end up like this. sage man on a computer and everyone with their faces down just looking at the computer to receive our education.

And I think we want, and I want to try to show to explain why.

Speaker background and motivation

And this is a special talk to me because it's really bringing together two lives, my present life as an AI practitioner at Mindstone with my past life as an education expert, mainly at the OECD here in Paris, which is an international organization where I spent five years giving advice to governments on how they should reform the education system.

And I think these two lives combined hopefully will allow me to tackle this question.

A classroom disrupted by AI

And I want to start from my own teaching experience. Because this year, beginning of the year, January, February, I got the chance to teach in a software engineering school here in Paris. It was really fun.

It was economics for first year students. first-year engineers.

I'm an economist by background. I had not looked at a lot of these concepts in a while, but I thought it was an interesting challenge.

And this is pretty much how I thought my experience was going to be.

Can you see the picture?

The “sage on a stage” myth vs. reality

So I thought I would be this sage man that, and these are AI generated, by the way, a sage man that spreads the knowledge about supply and demand to these eager students that are paying attention.

Of course, that was not the reality. This is a very sad, sage man with students just looking at laptops. This was my experience, at least for the first lecture.

And of course, I decided to do what I preach. So I faced the problem. and I decided to gather more evidence.

So I started walking around the room and I saw that the students were not on TikTok or WhatsApp, somewhere. But the greatest majority of these students was just double checking in real time what I was saying with all possible LLM tools.

And that left me wondering, what was, what is my role as a teacher here? What am I here to do? They have all the knowledge they can possibly need.

And they didn't even have 360 learning. So imagine if they had that.

Redefining the teacher’s role: from lecturer to coach

And so, I mean, I sort of understood that I really, and this was during the first lecture, that I really had to change my role as a teacher. And I didn't have to, my role was not to transfer knowledge, but first of all, to teach them how to use these AI tools. then how to take this theory and apply it to practice, and also to instill a sense of motivation that economics is important and they should take it seriously.

1So in a sense, I was moving from being a teacher to being a coach. It's almost as if I had a football team that already had the knowledge of how to play football, or a basketball team, choose your sport. And I had to sort of show them how to use it together also in a group as opposed to just give that knowledge to them.

A framework for learning in the age of AI

Now, I want to try to generalize this, try to explain what I think it means for the future of education at large. There are many ways of thinking about what we learn and why.

Knowledge, skills, and attitudes: key distinctions

I will take this framework, which is embedded in the cognitive psychology of the 50s and has been developed by different psychologists over the years, basically tries to make a distinction between knowledge, so things that we learn, knowledge, which are facts and concepts, skills, which is applying concepts to real problem, so rather than knowing how to do something, and then attitudes, which are dispositions and values that guide behavior, beliefs, mindsets, ethics.

What I want to try to do very quickly, because we only have 15 minutes here, is to understand how each of these learning dimensions is going to be impacted by AI. I mean, here are two examples.

Maybe let's go through one of the difference between knowledge, skills, and attitudes. What's the difference between knowledge, skills, and attitudes?

Maybe it's interesting to have a look at one. Let's think about economics, supply and demand, the two curves.

Knowledge would be, I know what happens as each of the curve moves. Skills, I can use that knowledge to actually understand what happens in the market. Attitudes, I believe in that tool of analysis as something that can be useful.

Bless you.

How AI changes access to knowledge

Now, knowledge, and we have seen it with Adrian, and to some extent with Benjamin as well. In a sense, having access to the knowledge is now trivial. The teacher is not the repository of knowledge anymore.

But what's important is that at least when you're using these skills, when you're using these tools without the mediation of a learning platform, like 365 Learning, that you can access it critically. So develop these critical thinking skills.

And by the way, here I don't mean that all six-year-olds should have access to AI models or should only do things through AI from the start.

I think a good analogy here is the calculator. We do not actually ask every single student, we do not let six-year-olds use a calculator straight away. We would pretty much ask them to do things by hand until the age of 10, 11, 12, depending on the education system.

Same things applies to the AI models. 1What we might need to understand is different cut-offs for different skills.

What happens to writing? What happens to math? So when it is that they can start writing essays with AI, when we think those skills are consolidated so that they can actually do the writing with AI.

But the point remains,

Evolving skills: cognitive vs. social-emotional

Now, skills is a bit of a tricky one because if you make this distinction between cognitive skills and social-emotional skills, so cognitive would be the production of knowledge or content, skills related to that. Social-emotional skills more understanding emotions, building relationships, reading the room.

A lot of, on the cognitive side of things, effectively many tasks will become doing things with AI, like the essay example that I've mentioned. So practical AI competence is very important.

But when we look at how I actually need to be in a room like I am right now and deliver a public lecture. I mean, I could be looking for tips with an AI model, but there is no substitute for the lived experience of actually going through the situation and having to apply that knowledge.

And so that does not, AI cannot really replace, teach these skills and replace teaching or education for those skills. They will need to be an element of practice. Maybe you can argue an AI avatar can tell you how you need to change behavior here and there, but you still need to be in the real world and you still need to practice that.

Attitudes and values remain fundamentally human

Attitudes, I feel is another area that becomes quite tricky because if you think about your beliefs, your mindsets, your attitudes to fundamental life problems, they are shaped over a long time. and they are not just going to be changed or they are rarely changed by the fact that you read some bit of text or you access just some course. And so this, the formation of these attitudes, I think will remain fundamentally human.

Now,

Synthesis and implications

Trying to wrap things up, and I know this was a tour de force, but I did want to try to give a sense of what I think the future would be in 15 minutes, which is always difficult. So, knowledge access is becoming almost complete, but critical thinking is necessary to ensure that we can make sense of this knowledge and we should really give priority to that across pretty much all the stages of the education journey, including with adults.

And then on the skill side of things, a lot of cognitive tasks will become AI first, but not attitudes and social emotional skills. where the role of education will still be critical.

Educators as coaches: modeling, guidance, and judgment

But all in all, this means that the educator is moving from teacher to coach. And so rather than just transferring the knowledge, as we thought about education for a long time, it's more about modeling standards, guiding judgment, and really teaching how to be and interact in the world.

Now, the pizzas have arrived, right? Yes. So I'm standing between you and the pizzas.

You can take as many questions as you want.

Live demo: Gamma and Claude in practice

I can also show you a little bit how to play with gamma, a bit more. I want to show you a feature that Benjamin did not manage to show you earlier. And maybe I can, as a bonus, I can also go through the Claude extension if you guys want. We'll do these two things very quickly.

Maybe for the questions on the talk, we do it over the pizza and drinks. It is what I would suggest. What do you think?

Real-time edits with Gamma’s agent

Now, one thing that this agent function is actually pretty cool. Because what you can do to this on gamma, so once you have created your presentation as Benjamin has done before, what you can do here, you can really ask to make changes on the spot. So if I go here, Can we make this a bit more visual, please?

And so this app, by the way, is called Super Whisper and allows me to dictate in every space or every field where I put the space bar. And you see that the standard is quite high. I'm not a native speaker, but I was able to understand what I was saying.

Hmm. all right what do we want let's try two not really sure if the please make a difference makes a difference but Some people say no. Yeah.

The tipping does, if you change your guidelines in ChartGPT. I can show you that.

You see, it did it. That was just on the back of a simple prompt. If we go back, it did make it more visual. And we can go back and forth between the two.

That's one thing I wanted to show you.

Connecting your apps: calendars, email, and notes

Now, on the cloud connection. So, I'm not sure you've seen this, but if you go here, search in tools, you see that you can add a bunch of different connectors. I already have my Gmail, my calendar, I can also add, for example, my note taker, which becomes really interesting because then I can say, take those notes and do something with them.

I don't need to copy and paste anything. I can build a project that basically says, make a proposal. And I can connect it if I follow the right steps. which should be fairly straightforward. So what I wanted to show you is how these connections make working with AI a lot easier.

So for example, can you give me an overview of my meetings for this week by looking through my calendar? And so now it's going through my calendar. Do you have a paid version or like a standard one? It is the team account. I think the connection should work regardless of the account. I think it's more the limits on the other features.

From meetings to action: summaries and follow-ups

So, here's a list of my meetings. Now once I have this, I can do pretty much anything with it. I can ask, prepare every single meeting, go through the list of attendees, and give me a summary. And for very important weeks where I have a lot of partnership meetings, I would do that. I would be able to ask for any information on how to optimize my time.

Yes. Sorry, I have a lot of questions. Could you, for example, if you have notes for a meeting that have already passed, say, go and take those notes from Firefly and give me something about it. Totally. Let's try.

Let's look for our meeting. I always need to find a meeting that has actually happened. Let's see. I had a meeting with Data Bird between two and three weeks ago. Can you retrieve information on that meeting from my calendar?

You see how the dictation really changes the interaction with the model? Because we are able to do a lot more quickly, because of course the alternative is to copy and paste prompts, but then I'm not able to improvise like we're doing now. Yes, I love when it works. By the way, it doesn't always work. I mean, it can fail.

No, that's now, this is the summary of the meeting. Yeah. The time is right. Yes. Now I've got this meeting. Now I can ask to find the notes.

So I think the note taker was that. Can you look through my fireflies for and extract some key points for this meeting? Now, this is where things become really interesting. Because really my cloud is becoming my operating system in a sense. It's bringing together all this patching information that is sitting a little bit anywhere. And obviously this takes a bit longer because it needs to go through the old transcripts on Fireflies.

I granted it. We have seen it when we went through the permission track. Did we have the note taker? It might be that we didn't have it. It might be that we did not have it.

let's see okay that is that let's see it's taking a bit longer than usual that might be why because i didn't allow it okay there you go you still need to understand the instructions okay it's a live demo things go wrong Yeah. Okay.

Imagine you're an experienced partnership manager. Can you write a follow-up email summarizing what needs to be done? Warm.

By the way, what becomes... Actually, don't use it that much for email follow-ups because I'm probably quicker. I tend to be quite generally a good drafter, at least in English. I use them for French.

Toward a personal operating system and digital twin

I'm generally a good drafter in English and probably quicker still if I do it on my own if I if I've just had the meeting Yeah, I think the interesting thing here is how bringing everything together really allows to have your own operating system and to work a lot more effectively now imagine when you start actually having a project and like I have that would have all your writing tasks loaded and is basically able to write like you. Now it becomes even more interesting because then you're sort of getting close to actually having this digital twin that can help you do things that you might not want to do.

Not replying to Benjamin, that was that idea myself. Yes?

Q&A and tooling preferences

What's the integration that you use the most from what's available today?

Fireflies. Fireflies. Yeah, Fireflies is useful.

Because have you seen that their search function is a bit fiddly? So this is.

For example, in there, there's Notion and stuff. But it's not still, it doesn't work quite yet.

Yeah, they introduced. They introduced it. Yeah, because I think it's a closer system.

I still need to play with it, honestly. I still need to look into it.

Usability notes: Claude vs. other integrations

So you can set it. Cloud is a lot more user-friendly. We were able to do it on the spot with, well, two clicks.

I got the second one wrong.