What’s wrong with the government’s AI strategy? Labour has been claiming today that it is ‘already out of date’, with shadow culture secretary Lucy Powell arguing that developers should be licensed by the government before they can work on advanced AI.

Powell has suggested that an arms-length body could run the licensing regime in the same way as medicines and the nuclear industry are governed. But when she pitched up on the World at One this afternoon, she didn’t sound fully up-to-date herself.

There needs to be more of a narrative from the opposition than just ‘we would do this better’

Sarah Montague asked quite reasonably why, if Labour was so worried about AI, the party hadn’t really been talking more about this before. She also asked whether the government was already doing what Powell was urging it to – Rishi Sunak is in Washington for talks with Joe Biden about precisely this topic, after all.

Powell insisted that the party saw lots of opportunities and risks in AI. She also said that it wasn’t fair to say the government was already there, because Sunak ‘is now quite late in the day coming to this: his own AI strategy was only published two months ago’.

When Montague asked how the regulatory framework might function, she replied: ‘I don’t have all the answers to that regulatory framework, but I would put in place very, very quickly a group of people working closely with the government to draw up some of those guard rails. I think we have got things that we can look at that are similar kind of models, when it comes to the development of nuclear power, when it comes to medicines licensing…’

The interview then deteriorated further when Powell was asked if she’d tried some of the popular AI tools such as ChatGPT or Bard. ‘I have tried it out of course, as have we all, but it does seem to have a safety mechanism in it that is not allowed to be political. So when I’ve asked for jokes about the Conservative party or something like that, it won’t produce those.’

She then added: ‘Well, I’ve just tried it out in different ways just to see how it might help me do my job more effectively, but it is remarkable how it can write articles and how it can do some of that work. I’ve also seen my teenagers try to use it for homework – unsuccessfully – but I think that’s just the tip of the iceberg, really, in terms of what it potentially could do.’

AI is moving fast, but Powell does have the luxury of opposition in which to learn a little about it, rather than just try to write jokes about the Conservative party. Although she’s the shadow culture, media and sport secretary who decided to complain that the Prime Minister is out-of-date on AI, it is perhaps unfair to single her out as being a bit inarticulate on AI: most MPs don’t seem to have a clue about it (they might even benefit from its writing abilities, given the quality of most speeches and questions in the House of Commons).

It has become the new 3D printing of politics: something largely tech-phobic politicians like to reference in speeches to show they understand the zeitgeist, without really knowing what it is or what it entails.

The problem for Labour is that Powell’s approach to AI is just one example of a wider issue that the party still has. It regularly attacks policies by saying ‘we would do this better’, without necessarily offering a proper analysis of the wider problems in a policy area, or indeed an explanation of why a Labour government would definitely be more competent.

Competence was one of Keir Starmer’s early obsessions, and given his predecessor and indeed the turmoil in the governing party, that was a sensible focus. But as the next election approaches, there needs to be more of a narrative from the opposition than just ‘we would do this better’.