It can help you get things done faster, but you can do serious damage if you aren’t careful.
AI is everywhere: in your web searches, the photos app on your phone, the maps you use to navigate, your email filters. It seems like every service is implementing AI-powered tools—especially large language models (LLMs)—and anyone who opposes its use is branded a luddite. Some people welcome its use, seeing it as the ace up their sleeve, while others worry that it will replace their jobs (or even threaten society as we know it). Part of what makes AI seem so threatening to the latter group is often a lack of understanding of what it is, how it works, or what it is meant to be used for.
At its core, AI is just another technology tool that can help people do their jobs more efficiently.
When AI was just entering the mainstream a few years back, I bumped into an old supervisor from my college landscaping job at the grocery store. We chatted for a bit, and when he learned that I worked in technology he immediately asked for my thoughts on AI, clearly worried. I had to think about it for a moment; what is a good way to explain the usefulness and dangers of AI in a way that a non-technical person would understand? My answer needed to be not just understandable and concise, but split the difference between complacency and catastrophizing. Here is what I came up with: “AI is like a gas-powered hedge clipper. It can help you get things done faster, but you can do damage or hurt yourself if you aren’t careful.”
While not everyone has used one of these (frankly terrifying) machines, nearly everyone has used a power tool. Most power tools don’t do anything that I couldn’t do with a hand-tool (or a rock, for that matter), but if you have ever done any DIY home repairs I am sure you would always reach for a drill over a manual screwdriver. The analogy to a power tool is, I think, a helpful one because it gets a few important concepts across that are abstract with AI but obvious with more familiar tools:
- AI, like power tools, can sometimes be overkill. I am sure we have all seen the “no power tools” icon in the IKEA furniture instructions. I am also sure that we all have ignored that warning immediately before stripping out or sheering off a cheap screw or fastener. Sometimes more power doesn’t mean a better result.
- AI can also hurt you if you aren’t careful. While the “hurt” that AI can cause is less palpable than the kind an improperly wielded power tool can inflict, careless use of AI can still cause serious damage. Reputations can be tarnished, trust undermined, and systems destroyed if AI is used without careful review of the final product it helped you develop.
- If you use it well, AI can save you time and energy. If I have to hang a shelf, I can get it done with a power drill and electric screwdriver WAY faster than with a hand drill and a screwdriver, but only if I take the time to measure twice, drill once, and double check I am using the right bits. AI can similarly save me tons of time if I know exactly what I can expect from it and know where it will not (and cannot) replace my know-how and experience.
- A tool is only as good as the person wielding it. Unless they are a mechanical genius (or read the instruction manual), most people will do a better job by hand than with an unfamiliar power tool. Similarly, there are many things that AI technically could do for you, but you will probably do a better job yourself than struggling to develop a prompt to do it for you.
- If you use power tools to work on something delicate, you run the risk of damaging that thing irreparably. Likewise, if you use AI on something sensitive or proprietary, you should be aware of the risk in using AI (especially open models). Some AI tools record what you put into them, and personal or private information should not be put into them. AI should be used cautiously for critical work where mistakes could be dire.
AI isn’t bad, but its use does carry a higher level of risk than doing things yourself. AI can help improve your work if the risks are understood and respected, but there is a learning curve to using it well. In short, just because AI is there doesn't mean it should always be used.