My GenAI "Huh" moment

Ever had a disruption epiphany? A moment where you experienced something, after which you leaned back and thought to yourself, “Huh; things are never going to be the same.” These are very powerful moments, and frequently they happen about things you were vaguely aware about, but didn’t fully realize their meaning until the ”Huh” moment hits.

?Here's an example: It’s the mid-90s, and I'm studying for my B.Sc. in Computer Science at 加拿大阿尔伯塔大学 (shout out). I’m working on an assignment for a networks class, for which I needed to get my hands on a compression standard by CCITT (now ITU).

So, what does a student who needs a professional document do? Head to the library, of course. I made my way to the Cameron Science & Engineering Library on campus, the source of all knowledge technical. However, to my amazement, and after several days of scouring the references, I could not find any book or document that contained the standard I needed.

That was puzzling, and also the first time something like this happened to me. If this huge library with its millions of titles didn’t have this standard, where could would find it?

OK, I reasoned, this is a communication standard, so perhaps a communication company would have it. I picked up the phone and called the offices of EdTel - our local phone company. Explaining what I need, I got transferred around until I reached the Engineering department, where a very helpful engineer took the call. “Sure,” he said. “I know that protocol, we should have the whole standard book here. Come on over, I’ll let you photocopy it.”

I hopped on a bus downtown to the corporate offices and met the engineer. We chatted a bit about my assignment, then took me to the company library. They had every type of communication standard ever published – except the one I was looking for. “I don’t understand, we’ve got to have it here somewhere,” he scratched his head. But after an hour-long search we had to give up, empty handed. It was at a dead end.

I returned home and slumped in front of my computer. I still had an assignment to do, and I’m the one who suggested this topic to my professor. Should I reach out to them and ask to change it? I didn’t want to admit defeat.

Then I remembered this conversation I had with some of my fellow students a few weeks before, about this new thing called “the web”. It was some sort of graphical version of the BBS and Usenet. My fellow students were very excited about it, although they couldn’t really make me understand what it was really good for. I also remember my boss at the computer lab I worked in showing me a web page showing an image of a fish tank that refreshed every five seconds. He was also very excited about it. I still didn’t get it.

I shrugged. What the heck, might as well try it out. I wasn't hopeful – I was looking for a very specific technical standard used in very specific use cases – why would it be out there in the world of slowly refreshing fish tanks? Still, I figured I had nothing to lose. Perhaps there was a clue somewhere out there on where I could find the standard? Perhaps I could get a lead on a library or communication company that had it.

?So I fired up my Centris 650, connected my dial-up modem, launched Mosaic, went on Infoseek and typed in the number for the standard. Immediately, I got back a link to the CCITT web page listing a bunch of standards, among them the one I was looking for. Near the number of each standard was a tiny icon of a document. I clicked on the one I needed. The standard document downloaded to my computer.

?I leaned back. "Huh," I said to myself. "I guess I don’t need a library anymore."

?It was an epiphany I needed to experience first-hand to fully understand. I mean, I knew that BBS existed, and that some universities put documents on it. Plus, digitization of documents was getting more popular – the computer lab I worked at had a scanner and I actually saw scan-to-text taking place. But it had not occurred to me that standard companies, instead of printing expensive books and distributing them via mail across the world, would just post them digitally online. And libraries were still the go-to resource for any research needs; it was inconceivable that something would replace them. Inconceivable, that is, until my own personal experience clarified to me just how easy it was going to be to do so.

??

Back to the present and to my latest "Huh" moment, this time with ChatGPT. Everyone’s talking about Generative AI, of course; nothing new here. But whereas many people view GenAI-based personal assistants as a threat to human jobs, my attitude towards it was was relaxed. I agreed with the oft-quoted saying, “AI won’t take a human’s job, a human working with AI will take the job of a human not using AI”. But I was thinking in the very specific context of a person working on a job; so it’s person in person out. Even if the person is more productive with AI than without, you just increase productivity with the same number of jobs. So no employee needs to lose a job because of AI. Or so I was thinking, until today, that is.

?Having recently lost a job myself (unrelated to AI), it was time for me to refresh my resume. I had a bunch of ideas and thoughts about how to reformat it and I wanted to consult a resume expert about it. Fortunately, part of the parting package from my old place of work included a free session with a job consultancy firm, and this is exactly the type of service they offer. Unfortunately, their availability is limited and it’ll be a few weeks before they’ll have an expert available to work with me.

Then I remembered ChatGPT. I been playing around with it for some time now, and one of the things I tried out was to adapt recipes I wanted to do – sometimes I wanted to change ingredients, or to make it more interesting, or to make a restricted version of it (like gluten free or vegan). I found it did pretty well in these cases, and specifically when I had back-and-forth conversations with it, sort of “thinking out loud” and trying things out with it. Could I do the same with my resume?

I shrugged. What the heck, might as well try it out. I wasn't hopeful – rewriting a resume is a much more complex task than tweaking a recipe. Still, I figured I had nothing to lose. Perhaps there would be a good tip from the general resume-building material that was probably used in its training that I could use.

?So I fired up my Mini, launched Safari, went to ChatGPT and uploaded my resume. Then I proceeded to work with it on taking my resume apart and rebuilding it. I adjusted, rearranged, and thought through various options together with it. I asked it about content, phrasing, arrangements, font sizes and more. I would also debate its replies, adjusting them to fit my own style. Some advice I took, some I didn’t (it leans towards formality and a bit of stiffness, which, I imagine, is how most of its training data is written), but it was very specific, gave examples, and made multiple good suggestions. And as always, it’s infinitely patient: I’d write something to it, get a reply, and several hours later think of a follow-up question. It answered immediately as if no time has passed. After a few hours spread out over a few days, I had my updated resume saved to my computer.

I leaned back. "Huh," I said to myself. "I guess I don’t need to consult with a human expert anymore."

Once again came the epiphany I needed to experience first-hand to fully understand. So it’s not just about an employee who doesn’t know how to use AI, who is under threat from an employee who does use AI. It’s about entire jobs that could potentially be supplanted altogether by AI. For most people, the sort of experience I had will suffice; they won’t need a human consultant. I’m sure some resume writing consultants will still be around, but clearly many of them will need to find some other work to do. And this is true for many other jobs involving reorganizing, realigning, or finding information. In short, this is the general-purpose AI expert.

?

Clearly, this is not any sort of new insight. People are talking about job loss to AI all the time – heck, I’ve been talking about it for several years now. But until now it was mostly theoretical; this is the first time I had an actual experience where I could see it taking effect. This is my “Huh” moment of it.

?

So now what? Well, I think we all need to check if our own jobs are in danger. Could an AI chat do what I do? It’s relatively easy to test – just take a use case from your work experience (one that doesn’t contain proprietary information or company IP) and try to do it as a non-expert using one of the AI chats. If you’re a technical writer, take a document you worked on for one of your clients and see if you could simulate them achieve your same corrections without you. If you’re a travel agent, take a complex request that customers made to you and see if you can simulate them getting the same results using AI without you. If you’re a financial advisor – you get the idea.

?Because if you succeed, then you need to think about what you do next. Better experience your “Huh” moment now, than have your customer experience it for you!

??

P.S.: I don’t need to tell you I used AI to proof this article, do I? Editors, want to check yourselves as well?