Preparing for an AI future

Rogue Robot is the first in a series of educational AI-supported digital immersive multi-user escape room experiences designed to prepare young people aged 16-23 for a future living and working with AI.

Explore six key components of the Rogue Robot experience:

1. Orientate

2. Investigate

3. Evaluate

4. Analyse

5. Diagnose

6. Debrief

Watch the orientation video and listen to Alice Smith’s call for help to understand your mission. 

Your team hurries to the Smith Family residence to see how you can help.

When you arrive, the place seems deserted… all except for the robot, which is racing around, waving its arms in the air!

Your search of the Smith Family residence will reveal several clues about the robot’s malfunction. But not every piece of information is going to help you solve the puzzle. Some items might even confuse matters. 

How do you prepare for a successful future career, if the jobs you’ll be doing don’t even exist yet? 

The answer lies in developing your most valuable skills. Your most human skills – that AI cannot replicate – like teamwork, creativity and critical thinking.

Once the team thinks they know what caused the robot to malfunction, it’s time to move on to the diagnosis.

Mission accomplished, it’s time to return to the Red Cell HQ for a debrief and to find out how the team got on.

According to the AI titans, human intelligence will soon be eclipsed by AI, but what do you think?

This chapter explores what we mean when we talk about artficial and human intelligences and challenges what the ‘architects of AI’ are overlooking when they talk about our ‘smarts’.

The term Artificial Intelligence (AI) is often applied to any software program designed to analyze patterns, automate complex jobs or generate content, but there are many different types of AI - some dating back as far as the 1950s.

Many AI systems are built using neural networks; a type of computing system loosely modeled on the organisation of the human brain.

Neuroscience often uses neural networks to try to model tasks performed by the human brain. However, it’s misleading to compare the workings of the dynamic, living biological brain with an artificial neural network, which is a greatly simplified mathematical model.

If the Silicon Valley CEOs are to be believed, AI is already smarter than humans in specific domains and will soon completely surpass total human intelligence. And yet, we’ve all seen examples of AI being remarkably stupid!

If you enjoyed this investigation into the relationships between human and artificial intelligence, here are some additional resources that you might also find interesting.

In Rogue Robot, people live in AI smart homes equipped with cameras, microphones and other equipment, which is used to monitor the occupants’ activities, health and wellbeing in order to anticipate their every need.

This chapter explores the implications of living and working with autonomous agents.

An AI agent is a software program that uses artificial intelligence to set goals, make decisions, and take actions to complete tasks on your behalf.

But is it safe to share your most sensitive and personal data with an AI agent?

In order to act as our personal assistants, anticipating and responding to our needs, AI agents need to monitor our movements and activities, as well as having access to our finances, email and other communications, medical details and so on.

This lesson asks whether the immediate personal benefits justify the loss of privacy?

If you enjoyed this investigation into the trade offs between convenience and loss of privacy, here are some additional resources that you might also find interesting.

Alignment problems are unintended and often undesirable consequences of giving AI instructions, objectives or training that is incomplete, ambiguous, or simply poorly thought-through.

This chapter explores why misalignment is the biggest challenge posed by artificial intelligence.

Ensuring that the AI models we create capture our norms and values, understand what we really want from them and, above all, do what we want them to do, is one of the most urgent questions in computer science.

AI isn’t coded or programmed, like other software is. Instead, developers train it to do what they want using reinforcement learning. But how do you teach something real-world goals, when it doesn’t live in the physical world?

There’s another big problem with AI alignment that we haven’t considered yet… 

AI alignment problems largely arise because humans are highly skilled at generalizing what we know to new, previously unknown, situations, while AI needs massive datasets to extrapolate information. 

Could the crux of the problem be that humans think in terms of the ‘why,’ while AI is more focused on the ‘what’?

If you enjoyed this investigation into the AI alignment problem, here are some additional resources that you might also find interesting.

For as long as humans have dreamt about artificial intelligence, they have also feared losing control of it, so why don’t we just give it an off-switch?

What happens if AI develops its own agenda and doesn’t want to be shut down?

This has been the subject of numerous movies and popular fiction and continues to be a hot topic today.

Corrigibility refers to AI system’s willingness to be turned off or allow its primary objective to be changed. 

This lesson asks what might happen if intelligent, autonomous machines realise that they cannot achieve their objectives if they are switched off. 

If you enjoyed exploring why an ‘off-switch’ is more complicated than it at first appears, here are some additional resources that you might also find interesting.