TECH TALK
Information and commentaries on Technologies and Science.

In a recent report published by the Secretary of State for Science, Innovation and Technology, the AI Opportunities Action Plan, the government lay out these wide ambitions for AI, with plans like building vast new data centres, nurturing a home-grown AI industry to rival Silicon Valley, and rolling out AI all across the public sector.

 

But with the government betting big on AI, are we going to be able to cash out? Or are we being sold down the river? Investing in AI seems smart, especially since it already helps many of us in our day-to-day lives. Who hasn’t used ChatGPT to help write an email, or solve some particularly tricky disagreement at the pub? In the tech world, Copilot, a model for writing code automatically, has also been making waves by drastically speeding up software

development. Models like these, which can read and write, are flagged in the report as having the potential to streamline a lot of civil service work, increasing productivity across the board.

 

AI could also revolutionise the other key areas identified in the report. Considering medicine, AI has already found use in detecting lung cancer from CT scans, identifying heart disease using ‘smart stethoscopes’ and forecasting A&E demand, amongst countless other examples. In education, AI has been touted to free up teachers’ time by automating marking, personalising teaching materials or even generating complete lessons from scratch. In policing, the National Police Chief’s Council have stated their ambition to be a global leader in AI, using AI to automate laborious parts of police work, or in a more sci-fi twist, predicting how likely criminals are to re-offend.

 

But this isn’t the full story. While it is powerful, AI can be very expensive due to its enormous computational requirements. For example, while ChatGPT has 400 million weekly users, its owners OpenAI are yet to turn a profit due to the huge cost in building, maintaining and powering the massive datacentres it needs to run. These costs

are already a blocker to widespread adoption of AI, and will continue to be without further technological breakthroughs.

​

AI also comes with other hidden challenges and risks. In healthcare, a recent report entitled ‘AI in London healthcare: The reality behind the hype’ found that AI tools provided to London hospitals rarely found their way

into regular usage. The same issue of cost was found by doctors, alongside practical challenges and a lack of trust in the AI. In education, concerns are being raised about the impact of AI on children, be it algorithms exposing them

to harmful social media content, or over-reliance on AI leading to cheating and impaired educational development.

In policing, AI can be just as harmful; AI policing technologies in the USA have been found to perpetuate racial biases, violate privacy and even lead to false incrimination.

​

While AI is already changing our world, we must be wary of any ‘miracle’ technology – real-world problems are difficult and rarely solved by any single technology, and the costs, dangers and practical challenges of AI must be mitigated for it to drive the kind of economic growth our government hopes for. So by all means, let’s put some

money on AI. But I think we better spread our bets.

​