johnmaconline

The first level of automation is Personal Automation.

Personal automation devices allow you to do your work faster, more accurately, and more effectively and do multiple things simultaneously. The electric drill and dishwasher are personal automation devices. Basically, they allow you to do something faster or more accurately and/or do something else while the machine performs the task.

Personal automation makes you more productive and generally happier. But you’re still intimately involved in the input and outcome. It’s like having to micromanage the shitty intern, but when properly directed, that intern is very helpful.

The drill hasn’t put any construction workers out of a job. No restaurant manager has fired the human who washes dishes because they bought a dishwasher machine. Both of these tools enhance those people’s contribution to the organization, but the organization isn’t materially benefitting beyond your uptick in productivity and happiness.

This is where AI, like ChatGPT, currently lives. You can offload tasks like editing your writing, creating ideas, summarizing TLDRs, creating code and script snippets, analyzing data, and generating images. It’s very helpful to you, the person, but you’re still required.

Like the electric drill, it’s not putting anybody out of work.

Next, we’ll look at the genre of automation that AI is starting to poke into — Partial Automation.

I see fear-mongering about AI all over the place. 

People worried about their jobs and their place in the world because of AI’s ability to automate. To be fair, some of this is warranted, i.e., AI automation will replace people in some roles. Including in my orbit directly. But let’s give it some context.

First, automating people out of jobs has been going on for 100’s of years through the introduction of machines into the workforce. Machines don’t get tired, have the strength of thousands of people, and can do the same thing repetitively with great precision. Nothing about the current wave of automation fear is novel or scarier than when machines joined the workforce. 

Second, as with the introduction of machines, certain activities and jobs are more exposed than others. The current fear-inducing AI, ChatGPT, and its competitors are a type of AI called “large language models.” 

“An LLM is akin to a vast library staffed by an incredibly fast and knowledgeable librarian. Just as this librarian can sift through thousands of books to provide information, summarize stories, or answer questions based on the content of those books, an LLM scans its extensive database of text from the internet to generate responses, make predictions, and understand context in a way that mimics human conversation and writing.”

The previous paragraph shows you what an LLM can do. ChatGPT wrote it. I asked it: “Describe what an LLM is in a couple sentences with an analogy.”

So, an LLM is good at certain types of writing. An LLM is also good at a few other things, such as generating images (and soon video) and writing code. 

So where does that leave you and me, and will AI automation put us out of a job?

As someone who works in both software development and automation (both squarely in the sights of AI), I have a unique perspective. I see three broad levels of automation:

• Personal automation
• Organizational augmentation
• Autonomous automation

Over the next several days we’ll take a look at these, what they mean, and how AI fits into each.

“We’ll just use AI for that!”

Heard that this week. And I heard something that sounded exactly like it in 1997: “We’ll just use the internet for that!”

It’s a great marketing pitch. It’s a great pitch from the CEO or the COO. It’s a great pitch for your social media feed or a TEDx talk.

But currently, just like in 1997 with the internet, that’s a meaningless, confusing, or possibly scary statement. It’s just buzzwords. Anybody can say it. Anybody can theorize ways to use AI. Most don’t really know what it means, what it can and can’t do, and how to make it do anything.

Which creates quite the opportunity for us humans. Just like in 1997.

A conversation I had recently:

Them: “I know you think that life’s better with AI, but I just can’t get on board with this AI thing.”

Me: “Really? Why not?”

Them: “AI will be our downfall. Our lives will eventually suck. That’s how they’ll get us.”

(Note: this person didn’t mean in the “Skynet-AI-will-kill-all-humans” way but in the “Vonnegut’s-Player-Piano” way.)

Me: “Ah, well, to be clear, my view is not ‘life is better with AI!’ My view is, ‘it’s an inevitable technology and it’s here and it will continute. Therefore, I want to learn about it, how to use it, and where the dangers really are.'”

Them: “But aren’t you afraid? Afraid that it will make you obsolete? Afraid that it’ll f— everything up?”

Me: “I don’t think about it like that because it’s here. I can’t change that. I accept it. I consider it like the car. Was life better before a car? Some think so. I could definitely argue that it was. But if you gotta get somewhere, would you rather have a car or not?”

Them: “Yeah, but nobody forces the car on me.”

Me: “OK, fair enough, but again, if you gotta get somewhere, would you rather choose a car? It’s a tool.”

Them: “Yeah, but I don’t understand AI. I don’t understand how it works. I don’t understand how to use it. I know how to use the car.”

And now we’re at the heart of the matter. The fear of the unknown. Which is all the more reason to dig in, figure it out, and join the conversation.

I do believe that life with AI has downsides. Maybe giant downsides. I think shared reality will suffer. I very much worry about bad actors and incentivization. I think truth may suffer. I’m concerned about the growing gap between have’s and have-not’s. I’m worried about more division. 

“Is life better with AI?” isn’t a very useful question at a macro level. AI is here. The cats out of the bag and ain’t going back in.

I think a better question is, “How do I make life better with AI?”

Seth Godin often laments the race to the bottom.

I’m with him. I paraphrase this as the hyperbolic version of “if some is good, more is better” — faster, bigger, cheaper, easier, less thought required, more dopamine engaged, bigger margins, all the colors, more buttons, etc. The Costco approach. “Why buy the amount I need when I can buy six times as much for four times the cost?!”

But another way I think of it is “lowest common denominator product and service development and offerings.” The race to the bottom is the illusion of “it’s better for everyone” — the public, the shareholders, the executives, etc. 

The paradox of the race to the bottom is that, taken over small windows and myopic variables, maximizing (or minimizing) can lead to a great product or service. Apple was a master at this in its heyday (which it has long since lost).

There’s no doubt that AI enables exponentially more. More videos created faster. More articles created faster. More books created faster. More web pages created faster. More ideas created faster. More choices created faster. More code created faster.

And more is better, right?