The modern debate about possible upsides and the potential downsides of Artificial Intelligence (AI) has been going on since at least 1990. That’s when Ray Kurzweil published “The Age of Intelligent Machines”.  However, the tempo has gone up a notch or three of late. This is partly because of sensationalist reporting in the media but also many of us have now had direct experience of using AI either by choice or, maybe more commonly, because we have been forced to interact with a chatbot. No human available.

Are all chatbots equal? No they are not.

The Large Language Models on which some of the better known AI tools have been constructed, including chatbots, have been shown to be riddled with error, not to mention the way they rip off creatives’ IP.

AI-based systems typically use natural language processing, but it still remains the case that if the data they are drawing on is biased, inaccurate, inconsistent or incomplete then the outputs will be biased, inaccurate, inconsistent or incomplete. This is just another way of saying “could be dangerous”. 

For me it was therefore highly refreshing to find a group of  people who were doing it right, using the power of AI to help children. They had constructed their tools using curated and carefully scrutinized sources.

I met them in Sofia.

Doing it right

In Sofia I was a speaker at a European conference coordinated by Terre des hommes Hungary.

Entitled “i-Access My Rights” it was the culmination of a project which explored whether or to what extent AI tools could be deployed in ways that would help children who were or  might become engaged with some or other part of the justice system. Well done Terre des hommes for having the vision, the energy and the courage to acquire the resources to put together this initiative.

Terre des hommes Hungary had a lot of help, in particular from its partners in

• Terre des hommes Romania
• Terre des hommes Greece and
• SAPI, a major social services agency in Bulgaria

They were ably assisted at a technical  level by

• KU Leuven’s Centre for IT&IP Law  based in Belgium
• Lawren.io a software developer, also based in Belgium, specialists in working with law firms
• Noldus  specialists in behavioural research, they are based in The Netherlands

All of the outputs were published in English but, by the time of the conference, many had also been translated and made available in Romanian, Bulgarian and Greek.

I believe the ambition is to make everything that is available in English also available in each of the other three languages. Perhaps other countries will be able to draw on and utilise the resources for their local use. I very much hope so. The production values were high which means the materials looked highly professional. The language was accessible throughout.

At the conference we learned that direct consultations with children and young people had been a key part of the project as it developed in its earliest stages then right the way through.

A great deal of clear headed thought had been given to the ethical and privacy issues raised and there was an acute understanding of the limitations of AI tools such as chatbots. But equally there was an appreciation of the challenges of finding scalable channels which would help children understand their rights when faced with having to engage with the justice system.

Doing a crude search in Google or Bing might lead a child astray. But if they find their way to Terre des hommes they can be certain they will be on  or will be directed to the right path.

In June in Madrid the issue of the value and role of AI tools will form part of a wider set of  global discussions about  how to promote fair and appropriate justices systems for and with all children worldwide.