Before we begin, here’s my experience with AI to give some context about me so you can know where I’m coming from:

• I use AI daily for work, probably more than most people. I’ve been building things with it since before the release of ChatGPT in 2022, including generating thousands of images (don’t ask).

• I’ve been coding professionally for three decades, as well as teaching university-level design and computer science for most of that time.

• I utilize AI in my role as a teacher, developer, and designer wherever it truly adds value. I encourage my students to use it too.

OK, here we go. These are seven of my own practice-informed opinions on AI.

1. AI is just a tool.

Only those who view other humans as tools believe AI will truly replace human work.

Worrying about ChatGPT in the classroom is no different than math teachers worrying about calculators in the 1970s. Asserting that our new “calculator for everything” will replace learning or teaching reveals a fundamental misunderstanding of both.

2. AI-generated results are mediocre at best.

Yes, Mid-journey, Sora, et al. generate technically OK images. But they’re easy to spot once you’ve seen a few of them. Frankly, they’re boring. No one is hanging these images on their walls1. At best, they serve as engagement bait on declining social networks.

AI music? Same. AI writing? Same. AI coding? Not quite the same.2

The mediocrity of generative output is both unsurprising and unlikely to change, considering that regression to the mean is literally how these models work. Predicting the average response to a prompt, with slight randomness and tuning mixed in, is unlikely to yield the next Mona Lisa without human intervention.

3. AI is not going to kill you.

There’s currently a large amount of FUD (Fear, Uncertainty, and Doubt) being spread across the internet by various parties regarding our current AI models. For example, there is a cult-like group of rationalist “researchers” who believe that AI is an existential threat as soon as AGI (Artificial General Intelligence) is reached. These “AI Doomers” are well-funded and prey on those who don’t understand how the technology works, and some of them are more dangerous than the thing they are warning against.

Most are not that dramatic, but instead spread partially-true FUD, such as the claim that ChatGPT uses bottles of water every time you use it (this is false btw 3).

It’s revealing that those working most actively with AI are not overly concerned about sci-fi-esque existential risks inherent in the technology itself, but rather focus on practical safety measures, such as preventing the spread of misinformation (aka FUD). Like any technology, irresponsible human use of AI poses the actual long-term risks to energy consumption and safety.

4. AI probably won’t take your job.

If AI does take your job, you may be happier for it—see Point 1 about tools. New technology is initially disruptive, but once integrated into existing systems, entirely new fields emerge. That’s where to put your energy. We still have math teachers, despite the fact that everyone has calculators in their pockets.

5. AI is overhyped.

AI has reached dot-com bubble levels of hype, and it will burst eventually. The gold rush is on but it won’t last. The right move is either to capitalize on the hype train or ignore it and continue doing what you’re doing (utilizing AI where it benefits you). The real long-term value will be found in subtler downstream aspects of transformative AI technology and will be developed by quieter, less hyped individuals.

6. AI is underhyped.

Remember the internet in the late 1990s? When the hype bubble burst in 2000, the actual practical applications unfolded in ways that were both profoundly impactful and almost impossible to predict. Here we are now on the other side of the hype cycle for the internet itself.

Today’s internet is a utility—like power or water—that we use daily without much thought or feeling..4 AI will be the same eventually. So keep calm, head to the Winchester, and be cautiously optimistic.

7. AI should not change how you live.

Stepping back to a more fundamental question: What are you working for? What are you trying to escape by letting AI do (or fearing it might do) your work for you? I’ll end with this poem by Joseph Fasano, because it says it better than I could:

For a Student Who Used AI to Write a Paper
by Joseph Fasano

Now I let it fall back
in the grasses.
I hear you. I know
this life is hard now.
I know your days are precious
on this earth.
But what are you trying
to be free of?
The living? The miraculous
task of it?
Love is for the ones who love the work.

1 It’s interesting that, since Midjourney v6, image models have generated technically better images than most human artists can. Yet we don’t value them at the same level as even a child’s drawing. There’s a parallel in chess, where AI has been technically superior to most humans for over a decade, yet human chess remains more popular than ever.

2 A human creator in any category can use AI effectively to create engaging and creative work, but that goes well beyond one-off prompting. Here’s an interesting recent example of someone who used AI to create a popular retro AI band. Regarding coding, the difference is that coding is fundamentally language-based, and plain English, or a close variation of it, may very well be the programming language of the future. See this talk by Andrej Karpathy: Software Is Changing (Again) for a more detailed explanation of where things seem to be going.

3 A single TikTok or Instagram Reel doesn’t stop costing resources after it’s uploaded. The data centre has to serve that file every time someone swipes it. Even a short clip (~5 MB) uses approximately 4 mL of cooling water per view. Multiply by 100k views and you’re easily at bathtub scale. By contrast, a ChatGPT prompt consumes about a third of a millilitre (e.g., a sip of water) once.

4 I wrote the initial draft of this post completely offline, sans AI; not by choice, but because the Wi-Fi was down at the coffee shop.