I think his product has a profound democratizing effect. Theoretically, a child living in a provincial town in rural Brazil should be able to get the same interaction with an Effecta AI teacher as someone living in Mayfair.
Will I lose anything by introducing artificial intelligence into the classroom? Will we end up with a generation of students who exploit chatbots as a crutch – to write essays, solve problems, and so on?
They will do it anyway. Trying to exclude artificial intelligence from schools is nonsense. It’s about how to incorporate artificial intelligence into education. Bad teachers will exploit it badly, and good teachers will exploit it very well – just like with blackboards and calculators.
But we are talking about a more fundamental change. I ask what it might mean that students are not developing basic skills.
If we go back to the time when calculators were invented, [people thought that] Children will never be able to do mental arithmetic. But that didn’t happen. This will obviously have an impact. However, I think the net effect should be positive in terms of educational outcomes.
Children are likely to be extremely vulnerable to the dangers of chatbots. How do you think about these threats?
Of course, there are dangers – vulnerable adults and children in particular become emotionally dependent and invested in a relationship with something that has a humanoid avatar in their lives.
At the societal level, we should take a very cautious approach. I believe that there should be a clear definition of the age at which agentic AI is made available to teenage people.
Do you like Australia’s social media ban for under-16s?
There is no point in introducing a ban if you cannot measure people’s age. This is where policymakers rush to grab headlines for bans and don’t fully consider some rather tough issues. Unless you want all these platforms to store everyone’s passport details? For a long time I believed that the only way to achieve this was to get through the iOS and Android bottlenecks [app store] level.
But in principle, I think you should take a similarly preventive approach. The susceptibility to emotional involvement and perhaps undue influence in your relationship with the kind, patient, 24/7 voice that listens to you all the time is very real.
However, I don’t think there is any risk in the case of the products Effecta produces.
Even though artificial intelligence is literally taking over the role of teacher?
Well, no – because it isn’t. These AI agents produced by companies like Effecta will not have some kind of secret midnight relationship where they say all sorts of spooky things to the student. It is a teacher-controlled experience.
You spent almost seven years at Meta. During this time, AI became the frontier technology. I’m curious how your experience at Meta has influenced your perspective on the possibilities, risks, and limitations of artificial intelligence and the search for superintelligence.
If you ask three people in the same organization what superintelligence is, you will get three different answers. I feel like everyone in Silicon Valley has to admit that they’re one step away from artificial general intelligence or superintelligence because that’s how they attract the best data scientists. It’s challenging for me to come to terms with such a twisted concept.