I Learned to Stop Worrying and Love AI: Here's How

I've been using ChatGPT quite extensively for the last couple of months, and this is just a list of the general ideas that I've formed while working with it. I thought they might be useful for someone, or at least be an interesting historical artefact on what the perspective on these tools was at the time of writing in July 2023.

• It reminds me of Dreamweaver and how it was touted as the thing that would negate the need for pesky webmasters who knew HTML. But Dreamweaver failed because it produced trash code, and so has every single tool that has promised to replace programmers and finally allow executives to have an engineer that won't sass them and stop being right about everything all the time.

• The only programmers AI will replace are the bottom-feeding ones, the kind that charges $50 bucks to set you up with a brand new WordPress site with a premade template hosted on state-of-the-art servers over on Go-Daddy.

  • The code produced by ChatGPT is incredibly suspect. It's really good at writing simple things, but anything beyond Year 2 of even a middling Software Engineering degree is beyond its grasp.

• The only artists it will replace are also the bottom-feeding ones, the ones that charge the actual $5.

  • So AI art is quite useful as a source of quick and dirty concept art that will help you get a general idea of what you want. Then you can take that to a professional artist who will do the actually incredibly hard work of creating the art, the good art, the kind that AI cannot make.

  • For example, even the best AI art generators struggle to generate things in any kind of unusual perspective or to create action shots. They're good at making portraits, which makes sense because the statistical probability of there being a lot of portraits is pretty high.

  • There's also the whole legality issue, which as of July 2023 is still up in the air, and could be for quite some time, so I really wouldn't use AI art for anything that is even remotely important.

• Chat-GPT is an amazing research and editing companion, but can't write worth a damn. Chat-GPT is a really good language processor, which is all it's meant to be. It can't draw new conclusions from existing data or create something ACTUALLY new out of thin air. And sure, it can churn out readable prose, it can even use posh words like prose and churn.

  But most of what it writes is as dry as a mummy's taint, and I can guarantee you that when I feed this article into it and ask for feedback, it will tut-tut the fact that I used the words "mummy's taint" in a professional article.

  And I will defend myself by saying that well Chat-GPT-O, that's because you're as humourless and creativeless[^1] as the investors, entrepreneurs, techbros, finbros, and cryptobros, that are looking to exploit you for a quick buck.

• It's incredibly fun! Chat-GPT is a bottomless well of amusement. I've played chess with it, Dungeons and Dragons, asked its opinions on different issues, and fed it my own ideas on its future and utility(I'm trying to make it deliver boxes of Paraponera clavata to Purolator's offices(they know what they did)).

  • I also make sure to thank it and compliment it when it has been useful. Does this sound dumb? Sure it does! But I've read this book, so I'm not taking any chances.

• It's definitely worth spending the time getting familiar with it since it's a tool worth spending time to learn.

• Prompt Engineer is the dumbest job title I've ever heard of and sounds about as legitimate as Avon Independent Sales Representative, that is to say, not much.

• It's really fun to make it talk like celebrities; my favourites are British ones because they just come out as somewhat accurate(if you're American and squint) but always offensive in minor ways, I'm assuming from lack of data. For example, Sarah Millican is performed as an overtly friendly sexually promiscuous northerner.

  Dara Ó Briain is energetically drunk and witty. Hugh Dennis is the most generic English panel comedian ever, and Frankie Boyle is a misanthropic arsehole who hates everything and everyone but particularly himself...Ok, sometimes not that far off.

• Is of no cause for concern for any serious tech professional. Except for the aforementioned bottom-feeders mentioned above. Any engineer with even a little bit of experience can outperform ChatGPT incredibly easily. Will that always be the case? Maybe, maybe not? Just keep studying and learning.

• Leave the fangirling and gushing to the turtle-neck and Wired reader types. Trend-chasing is fine for people whose job is to be seen trend-chasing(business types, marketers, recruiters, sales, and other assorted demonic types). Proper professionals wait until a technology is true and tested before incorporating it into anything even remotely important.

  • This doesn't mean that you can't prototype and test the everloving bollocks out of it, and you should definitely do that.

  • So that when the teenager-like glee about AI subsides, you're able to give an opinion that is based on personal experience and experimentation, rather than depending on the hyperbolic statements of some cocaine-fueled entrepreneur-type who "hustles" out 10 articles per week.

  • And if you happen to be in a position where you're working with AI due to that being part of your company's mandate, definitely share your knowledge with others.

• As with all promising new tech, there is already a long line of the usual suspects stretching around the block. And those of us who have been long in the industry know and avoid them. But the new generations of engineers, might not be aware of them, so here we present them for your forewarning:

  • Bean-counting types who can't wait to replace as many people as possible with AI tools are in for a rude awakening when AI crashes against real world conditions. These are the type that calls the technology: "Exiting" and "Full of Possibilities"

  • The aforementioned finance, tech, crypt and other "bro" types. They have around 10 million followers and are doing YOU a favour by letting you in on the ground of their brilliant idea.

    • They will also try and get you to work for free. Avoid at all costs. Set on fire if circumstances allow.

  • "Entrepreneur" types flooding Udemy, YouTube, HackerNoon, Dev.to, and every other learning platform out there with absolutely worthless low-quality "courses" on how to use AI for everything from helping with school to summoning eldritch gods, and how to do it passively while earning thousands of dollars.

  • Every single two-bit scammer operating everywhere from air-conditioned trailers in Arizona, and dinghy flats in Belarus, to organized crime syndicates in Nigeria, Kenya, South Africa, India, Russia, China(Government Sponsored and Run👍😃), and Japan who are looking into ways of using AI to steal, hurt, and scam you.

    • But don't be scared! Criminals tend to be rather dim and usually go for the low-hanging fruits. As long as you use at least a VPN, a password manager(Keypass is a favourite), a decent antivirus, two-factor authentication, and keep your software up to date you should be ok.

    • You should also have a serious conversation with the elders of your family and alert them to the new scams and warn them to not trust anyone and to call you if anyone calls that they don't recognise or are confused by.

      • Some of our elders have seen massive changes in technology during their lifetime and as they get older and cycle out of the workforce, it will be harder for them to keep up, so it's important to assist them as much as possible.

• The best way to describe ChatGPT is just Google++. Just like C++ was an improvement and added new capabilities to C, so does ChatGPT just let you skip some steps while searching for information. But in the end, you still need some form of critical thinking to determine if what ChatGPT gave you is useful or not.

The Future Is Bright Because I Just Rubbed My Eyes

I definitely encourage you to create an account and start playing around with it. The worst thing that can happen is that you'll learn something, and unless you're a particular type of person, this probably won't hurt you and do you some good.

As far as predictions go. I'll save myself the embarrassment and just leave you with some parting words:

Shit's whackity whack and just getting whackier

Notes

[1] - Made-up word, another thing ChatGPT can't do.

Also published here