AI Agents in Education: Why Cute AI Design Matters

We need to talk about how cute AI agents are becoming.

ChatGPT's newest feature, Dots, is described by OpenAI as "remarkably capable, always-on agents built to handle everything".

Let's sit with that description for a moment.

These are AI agents being positioned as capable of handling tasks for us. They can research, organise information, complete tasks and increasingly act on our behalf.

That's significant.

But one of the things that really caught my attention is how adorable the Dots are presented.

Why are we being given cute, cuddly characters to represent something that is becoming increasingly capable?

I think that's worth thinking about.

What are AI agents?

An AI chatbot generally responds when you ask it something.

An AI agent is designed to go further.

Rather than simply answering a question, an agent can potentially carry out tasks, interact with other tools and take action on your behalf.

That could mean researching something, organising information, making a booking or completing another task with less direct involvement from you.

This changes the conversation around AI literacy in schools.

We're no longer just talking about students asking AI to write an essay or explain a maths problem.

We're talking about systems that can increasingly do things for us.

And that means we need to think carefully about what we choose to hand over.

Why does it matter if AI looks cute?

There is nothing inherently wrong with making technology visually appealing.

But there is a difference between making a tool easy to use and making it feel like a little character or companion.

This is where AI anthropomorphism becomes interesting.

Anthropomorphism is when we give human characteristics to something that isn't human.

A name.

A face.

A personality.

A voice.

Language that makes the system sound like it has thoughts, feelings or intentions.

When an AI system is presented this way, it can become easier to forget what is actually happening underneath the design.

The cute little assistant isn't just sitting there being cute.

It may be processing information, accessing data and carrying out tasks on your behalf.

The design can make that feel much less significant.

What are we actually outsourcing to AI?

Imagine your AI assistant can research something for you.

That's convenient.

Now imagine it can book your travel.

Also convenient.

It can remember information about you.

It can use your data to personalise its responses.

It can potentially interact with other services.

Each individual action might seem fairly harmless.

But put them together and you have a system that can become deeply embedded in how you work and make decisions.

That is why I think we need to be deliberate about what we choose to outsource to AI.

The question isn't simply whether the AI can do something.

It's whether we should let it.

Convenience can make us stop questioning

One of the biggest challenges with increasingly capable AI is that convenience can be incredibly persuasive.

If something saves us time, we naturally want to use it.

But there are some things we should pause before handing over.

Research is one example.

If AI does all of the research for us, we may lose the opportunity to develop our own research skills.

Decision-making is another.

If we increasingly ask AI to compare our options and tell us what to do, we may gradually become less comfortable making those decisions ourselves.

The same applies to students.

If an AI agent can handle a task from beginning to end, students need to understand whether completing that task themselves is part of what they are supposed to learn.

That is a core part of responsible AI use.

Why AI literacy needs to include design

AI literacy is often discussed in terms of technical skills.

How to prompt.

How to use different tools.

How to get better results.

Those things can be useful.

But AI literacy also means understanding how technology influences us.

Students need to learn that the way a technology is designed can influence how they interact with it.

A friendly interface can make something feel trustworthy.

A human-like voice can make an interaction feel more personal.

A cute character can make a powerful system feel harmless.

None of those design choices tell us whether the technology is actually safe, accurate or appropriate.

Students need to be able to separate the design of a product from the capabilities of the technology.

The UK's approach to anthropomorphised AI in education

This isn't just something I have been thinking about.

The UK's product safety standards for generative AI used in education settings specifically address anthropomorphism.

The guidance says educational AI products should not be anthropomorphised in ways that could encourage children to believe the AI is human or alive. It includes considerations around names, characters, personalities and language that suggests human characteristics.

That is a really interesting approach.

Because if we accept that design can influence how children understand and interact with AI, then the way we present these systems matters.

Maybe adults need some of that thinking too.

What does this mean for schools?

Schools don't need to be scared of every new AI product.

But they do need to become more thoughtful about how these tools are introduced.

When considering an AI tool, schools can ask:

What can the AI actually do?

Look beyond the interface.

What information can it access?

What actions can it take?

What systems can it connect to?

What data does it use?

Understand what information students or staff are putting into the system and how that information is handled.

Does the design make the AI seem more human than it is?

Consider whether the tool uses characters, personalities or language that could encourage students to treat it like a person.

What are students being encouraged to outsource?

This is perhaps the biggest question.

Is the tool helping students develop a skill?

Or is it simply removing the need for them to practise it?

We need to stay deliberate about AI

The more capable AI becomes, the more important it is that we stay deliberate about how we use it.

AI agents may be able to handle more and more tasks for us.

That doesn't mean we should hand everything over.

And making an AI system cute doesn't change what it is capable of doing.

In fact, the cuteness may be exactly why we need to pay closer attention.

Students need to understand that AI isn't their cute little assistant simply because it has been designed to look or sound that way.

It is a technology with capabilities, limitations, data requirements and potential consequences.

That's why AI literacy needs to go beyond prompting.

We need to help young people understand not just what AI can do, but what they should choose to let it do.

And perhaps, just maybe, adults need a little more of that education too.