Nick Clegg Prefers Not to Discuss Superintelligence

I believe the product has a significant democratizing impact. Ideally, a child in a remote town in rural Brazil should experience the same interactive engagement with the Efekta AI teacher as a student residing in Mayfair.

Are there any drawbacks to introducing AI into classrooms? Will we create a generation of students who rely on chatbots to compose essays, solve problems, and more?

They’ll do that regardless. Attempting to exclude AI from educational settings is futile. The focus should be on how AI is integrated into teaching. Ineffectual teachers will misuse it, while effective educators will leverage it efficiently—just as they did with whiteboards and calculators.

However, we are discussing a more profound transformation. What are the implications for students if they fail to cultivate fundamental skills?

Historically, when calculators were first introduced, there was a belief that students would lose the ability to perform mental calculations. This, however, proved to be unfounded. While there will certainly be impacts, I believe the overall effect on educational outcomes should be beneficial.

Children may be especially susceptible to the risks linked to chatbots. What are your thoughts on these dangers?

Indeed, there are risks—especially concerning vulnerable adults and children who might become emotionally reliant and invested in interacting with a digital presence that resembles a human.

On a societal level, we ought to adopt a cautionary stance. It’s essential to establish clear age restrictions regarding access to autonomous AIs for young people.

Similar to Australia’s social media restrictions for those under 16?

Implementing a ban is futile if effective age verification isn’t in place. Policymakers often hastily pursue headlines about such bans without considering the complex realities. Unless platforms intend to require personal identification data, my long-held view is that this must be managed through the control points of iOS and Android at the [app store] level.

In principle, I support a similarly cautious approach. The danger of becoming deeply emotionally attached to and potentially influenced by a gentle, attentive, always-listening voice is indeed significant.

However, I don’t perceive any risk associated with the types of products that Efekta develops.

Even though the AI takes on the role of a teacher?

Not exactly—because it doesn’t. The agentic AIs created by companies like Efekta will not engage in any covert midnight conversations relaying harmful messages to students. It is a teacher-directed environment.

You spent nearly seven years at Meta. During that timeframe, AI became the cutting-edge technology. How did your experience at Meta shape your views on the opportunities, risks, and limitations of AI—and the pursuit of superintelligence?

If you ask three individuals from the same organization to define superintelligence, you will likely receive three distinct interpretations. It seems everyone in Silicon Valley feels the need to claim they’re on the verge of achieving artificial general intelligence or superintelligence to attract top data scientists. I struggle to engage with a notion as nebulous as that.