The cost of AI is someone else's time

I have a love-hate relationship with Large Language Models. At first, I'm like a kid assisting a magic show. I remember how amazed I was when I saw the first tech demo for ChatGPT Omni and its cough definitely-not-Scarlet-Johansson's voice. The movie “Her” was getting real. Yet, besides some very specific use case like using LLMs as the new Google for everyday troubleshooting, I'm frustrated by AI more often than I'm delighted with it. This is true when I use it, but more importantly when it's thrown at my face, unsolicited.

The genie is out of the bottle and GenAI is here to stay. What we can influence is the etiquette around its usage, especially in close social circles and at work.

Below is my compendium of AI hot takes and rules of thumb I wish everyone I interact with would keep in mind.

If you feel the need to specify your answer is AI based, work it more

I see this becoming a trend in the corporate world, you send a request to a colleague, and they answer with a gigantic wall of text, word document, or tabular content. As a preamble, you have a one-liner mentioning this was done “with the help of AI”, as if this would absolve themselves of their responsibility if the content was bad. More often that not, they’ve barely read, let alone reviewed the produced content.

This breaks the trust relationship as I can’t tell if I’m reading your judgment or the model's. If I asked you in the first place, it was to get YOUR opinion or approval, based on YOUR expertise. I can prompt GenAI tools just fine on my own.

Feel free to leverage LLMs during your work, but you need to own your work fully. Guide the models with your expertise, proofread their answers, amend where needed until you reach a point where you’re proud enough of the result to put your name under it.

If you are incapable of assessing the quality of the model's output in a specific domain, do not rely on it

LLMs make for incredible semantic search engines and summary generators, but they can also be blatantly wrong while writing in a very confident tone. We call them hallucinations and even the best models available still do this regularly. This is a problem that might never get solved.

Hallucination is actually one of the main reasons I’m often deeply disappointed by LLM in real use case. I would ask for advice on how some construction work, only to realise later it recommended me to put on walls a product that is only designed to be used on floors, etc. It’s destabilising as most of the computer science world, its governing algorithms, used to be about deterministic behaviour. You would ask a program something, and it would always give you the exact same answer. LLMs however, are stochastic beasts. Their output can vary wildly, yet they mimic human behaviour and answer with confidence.

This is a dramatic difference with human-based interactions. When people have doubt, don’t recall exactly, or don’t know, they will usually mention it outright or refrain from giving any feedback. They won’t give you a made up answer only to say “Yes, you’re absolutely right!” when you confront them about their bullshit. People behaving like this would quickly lose credibility and get ignored or ostracized. Paradoxically we have an extremely high tolerance for this behaviour with LLMs.

If you use LLMs on areas you’re not comfortable with, be overly sceptic of everything it says, even when it confirms your natural bias. Cross-check whatever you can with other sources or at least double-check the LLM’s own source when it’s the result of a web-search and the website are referenced.

If you didn’t put effort in writing your piece, I won’t put effort in reviewing it

For the first time in history, producing content can be faster than consuming it.

LLM tend to be overly verbose by default. They will produce gigantic walls of text, neatly organised with titles, bullet points, bold highlights and emojis. This leads to a gradual inflation of the average corporate email, ticket, PR, or specification document. I’ve seen people use this as a weapon, bombing you with a 10 pages long AI generated pamphlet pushing against something you’re trying to push and asking you to argue back with the same level of details. Those pieces of content tend to be of very low entropy, abusing rhetorical constructions and repeating the same points over and over; two paragraphs’ worth of ideas diluted in 20 paragraphs of text soup. This creates a strong asymmetry in the relationship where the receiver has to spend a considerable amount of time reviewing what only took a fraction of this time to produce. Alternatively, they can give up, shove it to their own LLM of choice, and ask to generate a summary or directly an answer. At this point, you’re basically doing an GenAI Ouroboros, a total waste of time.

The only sane way to fight back is to plainly refuse to take part in such madness and ask for a more refined document, going straight to the point and properly vetted. This can of course be politically challenging depending on who you’re interacting with. An easy way to defuse push-back is to politely ask them to provide the prompt they use to generate it. At best, you’ll get a messy but much denser explanation of what they actually wanted to say. At worse, you will get a one-liner as the smoking gun showing their lack of engagement.

If your prompt is not at least half the size of the content you produced, it probably needs more context

The “Thinking” models are making this point less relevant, but LLM are, in essence about generating the most plausible next token based on the existing prompt or past conversation. If you give them very little information to chew on, they will give you the most average, one-size-fits-all soup possible. Working with LLM can be frustrating because, contrary to a colleague, even a junior one, they never “learn”, they will keep forgetting about context, past mistakes and other lessons learned. Some supporting software try to address this by adding “memory” capacity or grabbing context automatically based on connected apps, but the outcome is way more inconsistent than what the average human colleague would do. The more you steer the model by pouring your own expertise in it, the better the outcome. This is of course dramatically reducing the time saving aspect of using LLMs in your work, but it is the price to pay to avoid mediocrity.

I’m sure we could argue forever on what is the sweet spot for input/output ratio, but you get the idea. To be transparent, the situation is generally improving. A year ago I would have recommended for your input to be about as long as the output to get acceptable quality.

Any skill you heavily delegate to GenAI will atrophy.

The brain is not a muscle, but it definitely behaves like one. You cannot outsource the pain of learning, you have to challenge your brain and fail until you succeed to improve. Acquired knowledge doesn’t last forever either, the brain progressively purges unused skills and wisdom, the same way an unused muscle atrophies.

There is growing empirical evidence that heavy reliance on AI for cognitive tasks can lead to declining abilities but also reduced engagement with the underlying work. The scientific literature on this topic is still young but tends to confirm this. We’ve seen this pattern before. The average person is worse at mental calculus now that calculator is widespread, the same goes with orienteering and the extensive use of GPS.

This is ironically used as a counterargument for people in favour of expanding LLMs’ usage. It goes like this: “We’ve been outsourcing to technology for decades, and we are not a lesser being because of it. What we delegated to the machine gave space for our limited brain time to focus on”. A common counterargument is Plato's criticism of writing itself: books would weaken memory because knowledge would be stored on paper rather than in our heads.

It’s not that simple, LLMs don’t outsource a specific, narrow task. The applicable use cases are so vast, it allows us to delegate essential skills like planning, reasoning, decomposing problems, decision-making, or even creativity work.

Short terms, you’re definitely saving time dumping your unorganised, messy chain of thoughts into the prompt and asking the LLM to structure it into a neatly written email. But this is at the cost of progressively getting worse at doing it yourself, to the point where you might lose your capacity to make an articulated argument on your own.

I’m not saying we shouldn’t use AI, but it’s important to be well aware of the consequences. Keep delegating the meaningless work who doesn’t leverage your skills. For what matters, instead of asking the LLM to produce the content for you, use it as a sparring partner to challenge your own production. You’ll trade speed for quality, and you'll keep leveraging, sharpening your own abilities. By challenging the LLM feedback and not taking it like a ubiquitous oracle, you’re also ensuring the produced content keeps the style that make it yours. Then you also avoid the awful GenAI soup full of distinct biases, like em dashes, short and punchy sentences, and other cheap rhetoric phrasing from plaguing your texts.

Own your work

All of the above can be summarised in this golden rule: Do what it takes to stay proud of what you produce.

If you use AI, the responsibility for the result remains yours. Always verify, refine and challenge it. Remove everything that doesn’t deserve your name under it.

Respect people's time, they'll respect you in return.