"Five Government departments are using artificial intelligence to answer questions in Parliament.
Labour ministers were accused of providing "stock answers generated by a computer" when responding to questions tabled by their fellow MPs.
It comes after it emerged that Peter Kyle, the Science Secretary, has used AI software ChatGPT to help him come up with policy advice.
All departments were asked last month by Tory backbencher Peter Bedford whether they were using AI to respond to MPs' parliamentary written questions.
The Department for Business and Trade, the Department for Education, the Department for Work and Pensions, the Department for Science, Innovation and Technology and the Foreign Office all confirmed that they were."
The Daily Telegraph reveals
This is a good thing because it speeds up the required research, it formats intermediate output in a way that is easy to follow - use of headings and bullet points etcetera - and it helps write quickly in a convincing, but also standard style.
Would someone critical of this use of Al also recommend that the prime minister travel to work by horse-drawn carriage rather than EV?
With the right series of prompts, AI will format output that can be checked and then fed back in to write a suitably nuanced answer in a style suitable for the houses of parliament.
For a writer it's most helpful., it helps collect and analyse and organise and support thoughts into a broad and balanced comprehensive perspective of the writer's choosing. This output can b returned for help in communicating an interesting output.
So far so good for PMQs ?
Well, good for the writer, looking for a speedy, comprehensive and stylish text. But the writer is not making policy!
The RASCI framework is widely used in organisations for deciding the roles of participants in a project.
If you "think RASCI", it's a way for assigning responsibility in a project team, you can immediately see that the minister answering questions is accountable to parliament for government policy, but is not responsible. Accountable but not Responsible. You could say that his Support team and simply Informs him of what he must Communicate.
Shock horror?
Not really because it's always been that way. All that's happening is AI is helping the team of civil servants produce standard answers more quickly. If the minister's team is responsible for the content, then presumably they Validate and Improve the AI output, before giving the minister their "lines", and the minister Verifies before delivering to parliament.
The risk is the same everywhere, which is that with this massive and powerful support, we become lazy and uncritical, our mental faculties decline, and without a strict error check and validation process, for the minister to sign off, then things could go very wrong. Because surprise surprise AI can make mistakes.
0 comments:
Post a Comment
Keep it clean, keep it lean