Did you catch Merriam-Webster's word of the year last year? It was “AI Slop”. According to their definition, it is “digital content of low quality that is produced usually in quantity by means of artificial intelligence.”
Now, this isn't the same as AI revulsion. This is different – it's a strong, negative reaction to work that has clearly not been vetted. Someone prompts AI, copies the output, and sends it along. AI work can be done well, but usually the output of AI is middling. It's fine.
But it needs your big, beautiful brain to amplify it.
A lot of the conversation is about how “AI can amplify things you do” – I think that is 180 degrees from the actual truth – we amplify AI. AI can get us in the ballpark, but we are the ones that put the finishing touches on it. We are the ones that make the product spectacular.
The reaction to AI slop is real, by the way. Harvard Business Review published their findings about work slop and it is even more abysmal than you might think: half of the colleagues who observe work slop from an individual view them as less creative, capable, and reliable. Forty-two percent saw them as less trustworthy, and thirty-seven percent saw them as less intelligent!
A new term has emerged for people who pass off AI slop – Meat Proxies. Humans who just ship whatever AI spits out.
This is not an indictment on AI – in fact, when used well, AI can really potentiate what you do! Here's a small framework that you might want to add to your decision matrix when using AI:
Deputize – Use this when the stakes are low. Prompt once and you're done. You expect that the output will be just fine, but that's okay; you accept the risk. Maybe you need to write an email to ask your neighbor to borrow a ladder. Or you are planning a workshop and ask AI to generate ten icebreaker questions (if any are duds, you skip them).
Delegate – Use this when the stakes are moderate. The output is something you intend to buff, but it isn't mission-critical. AI drafts, you edit and approve. Maybe you ask AI to write product descriptions for the items in your online shop and you fix the details and add your voice. Or you send an email to your class about an upcoming deadline – if there's a mistake and you don't catch it when reviewing, it's easy to send a follow-up.
Duet – Use this when stakes are high! You and AI are co-creating. AI is a second set of eyes; a sparring partner that pushes your thinking. You are the expert, bringing your vision, judgement, and nuance. Together you arrive at something you could not have done on your own – and neither could AI. But here's the thing – your ideas may be entangled with output from AI, but the accountability is all you. Giving a keynote talk? Writing a grant proposal? Pitching a business plan? Vision and mission statements? The fallout could be reputationally damaging if you aren't vetting every step of the way.
Don't – Sometimes the right amount of AI is none. Don't use AI when the doing is the point, when the human connection is the point, when you can't verify the output, or if it deprives you of an opportunity. I don't use AI when I'm making slides because I love making presentations. I don't often use it for writing because I like writing. I would never use it for writing a condolence card for a friend. I tell my learners not to use AI for their first few lab reports (the struggle and friction are where learning happens). And I certainly wouldn't put personally identifiable information into an AI tool.
If you missed the webinar on AI Slop that the FLX AI Hub produced a few weeks ago, check it out! The biggest call to action, though, is for you to develop your own decision matrix for when - and how - to use AI. Think about the reputational risk, think about the blast radius, think about the joy and passion you need to retain, and think about what you'll miss if you are hands off.