AI moves into companies faster than oversight can keep up

This week paints a picture of artificial intelligence having truly moved into companies' engine rooms, but with governance lagging behind. Large companies are committing to AI for years, while others struggle to get an overview of what they have already let in.

Novo Nordisk is one of the clearest examples. The Danish drugmaker wants to use the chatbot Claude to find new drugs faster. At the same time, several technology companies are building their AI assistants directly into the systems employees already use for sales and customer service, so the AI can act without you having to switch between programs.

But the deeper AI digs into companies, the harder it becomes to keep track of. A new survey shows that half of Danish companies don't know how employees use AI day to day. The same pattern can be seen on the money side. Because AI requires special computing power, cloud providers such as AWS and Google send unpredictable bills, and more IT managers now have to renegotiate their agreements to get costs under control. Even OpenAI acknowledges that more ChatGPT users are not in themselves proof that the AI is creating real value for the company.

Fewer hires with AI

This raises a bigger question about what AI actually does to the labor market. Danmarks Nationalbank has looked at Danish companies and finds that those using AI hire fewer new employees than comparable companies without AI. These are the first concrete Danish figures on a development many have predicted but few have been able to document.

Dispute over the pace

At the same time, another debate is raging about how fast development should go. Meta CEO Mark Zuckerberg clearly refuses to put the industry on pause and believes each company is responsible for developing safely. The Chinese state media outlet Global Times goes further and calls calls for a slowdown a "Cold War tactic" against China. Right now, the race is probably more about rhetoric than real brakes.

Overall, the week points in two directions at once. Companies are committing more closely to AI than ever, but they still lack answers about what it costs, who uses it for what, and what it does to jobs.

It is therefore worth spending the coming period getting answers to three questions internally: Do you know which AI tools employees actually use? Does your cloud bill match your actual AI usage? And have you considered whether AI is already affecting your hiring plans, as Danmarks Nationalbank's figures suggest for Danish companies in general?