Should ministers use ChatGPT? The Dos and Don’ts of using Gen AI at work.

Secure and more reliable versions of Generative AI tools are finally rolling out into the hands of employees everywhere. Even UK Government ministers. So it’s time to ask a big moral question? When and where should you use it. And where shouldn’t you?

Do

1.??????? Be safe. Make sure you know where your uploads and questions might end up. If the tool you’re using isn’t ‘locked down’ and localised then don’t upload your anything you wouldn’t want someone else to see. If in doubt, ask.

2.??????? Check everything. These tools aren’t yet faultless. They were designed to guess the next word or dot. They were designed to generate. So they do make mistakes.

3.??????? Check sources. Generative search (think Google’s AI Overview or CoPilot powered by Bing) summarises websites so you don’t have. Great, until that summary is pulled from an unreliable source. Always ask for sources and check them.

4.??????? Make edits, especially if you use British English. Most GPTs spit out US English. And take out the unnecessary bolds and emojis, unless you really love them. I don’t.

5.??????? Be transparent. It’s ok to use Gen AI for content creation and ideation. In fact, it’s even smart because you’ll save time and come up with some really brilliant outcomes. But it’s so important to disclose where and when you have so that colleagues can check your thinking and run any legal checks required.

6.??????? Experiment. I’ve been using LLMs for over two years and I’m still blown away everyday by what these tools can do. You’ll find yourself saving hours and hours of repetitive task time every single day, leaving you more time to do the things that really add value to your business.

7.??????? Stay patient. They can’t do everything. Yet. They will get things wrong, but that’s not the time to give up. Those that keep persevering will be in a stronger position in the long term.

?

Don’t

1.??????? Give up. If I had a pound for every time I heard a colleague say “It got that wrong” and then gave up using a tool, I’d have perhaps ￡93.? But that’s 93 colleagues who should try harder. The tools learn from your conversations with them. They are not designed to be transactional. And they’re getting updates to make them better. So giving up is like giving up on the intern who’s desperate to learn. You wouldn’t do it.

2.??????? Write blog posts entirely with Gen AI – unless that’s the capability point your trying very publicly to make. Blog posts are supposed to be your opinion, not a borrowed opinion from millions of other data points. Just as you wouldn’t claim to create a piece of AI art or music, if you haven’t written the piece yourself, don’t claim you have. Don’t be afraid to keep being yourself. Guiding the machines in the direction we want them to travel. In time, when the machines learn more about us, this will change. But not yet.

3.??????? Replace human judgement. If it looks wrong, technically or morally, it probably is. GenAI isn’t designed to replace humans, it’s designed to be incorporated into workflows, overseen by humans. It’s therefore important to keep yourself within that loop, making decisions all the way through it.

4.??????? Overuse it at a personal level. Imagine how you’d feel if every digital interaction you had with a colleague began to feel machine generated. You might feel like you’d lost that human relationship, which is so important in business. So think carefully before writing every email with generative prompts. It will be harder to do but it will be worth it in the long run.

5.??????? Don’t claim huge productivity uplifts overnight without crediting how you did it. You’ll only get found out later when everyone else catches up. Sharing is really caring here. Your business will only benefit if you bring colleagues along with you as you make and share great productivity discoveries.

6.??????? Don’t trust platforms and tools you don’t know. Use business-approved platforms for business tasks. This is still a very new technology and few really understand where – in an open environment - all of this knowledge ends up. Going right back to the top, if in doubt, ask. ?

GenerativeAI didn’t write this blog piece but I did use it for my research.

Disagree? Anything to add. Would love your thoughts.