Is Your Job Threatened by AI? The Off-Ramp for All the Jobs That Will “No Longer Exist in 6 Months”

My favorite AI tools are months-old, and they’re already taking our jobs. You should have seen ME at eight months: unemployed, with no communication skills, and contributing nothing to the family business. I’m not gonna lie, I was a bawling mess at their age.

Midjourney, ChatGPT, and Dall-E are spectacular platforms that will take your breath (and your white-collar job) away. With job automation hitting some of our best jobs, the big topic in society is “What will you do in 6 months?”.

A closer look at what we’re competing against

AIs are not replacing redundant labor and the jobs nobody wanted. They completely disrupt the aspirational jobs, the “what do you want to be when you grow up” kind. AI seems to replicate skills we thought were unique to us humans: creativity, problem-solving, connecting dots.

AI replicates creative thinking. AI is great at noticing and replicating the patterns it sees. There is a sense of creativity emerging out of these patterns. I’ve seen the AI toy with visual symbols it doesn’t begin to understand and nailing it.

Even a broken clock shows the correct time twice a day. AIs iterate so much, at such great speeds, it almost doesn’t matter when they’re missing the mark.

AI is fast and efficient.  Let’s say your AI speeds up your work by x%. Ideally, the sales department should sell x% more company time to clients. Otherwise, you can safely assume that precise percentage of people will be laid off. Speed is not something you can compete on, just like you don’t compete on head hardness with a hammer.

There are AI tools out there that genuinely make work 80% faster. There is no way demand can keep up with this level of job automation. Creating regulations for a four-day workweek will have as much effect as throwing a glass of water on a house fire.

AI connects the dots. ChatGPT started on a dataset of 45 terabytes of text. This initial dataset trained it to understand the structure and patterns of language and generate coherent responses. It also feeds ChatGPT a world of valuable information.

AI’s capabilities are well beyond searching through content. AIs can use their data to make informed decisions. They look for greater patterns that only emerge at scale. They sort through an incomprehensible amount of text and weigh in relevance.

If you’re an avid reader, you’re working with less than 2 gigabytes of books and probably can’t remember the middle name of Holden Caulfield. There’s no amount of energizers we can drink to match that speed and scale.

A closer look at what AI tools are competing against

We never truly matched our tools in speed and efficiency and power. My car can outspeed me, and my calculator can outcalculate me. There’s nothing intimidating about that.

Artificial Intelligence has its monkey-see, monkey-do capabilities. It’s easy to imagine how, just like a baby, it could eventually stop imitating what adults do and develop its own process. Looking at the tech behind it, I’m not sure we are that close to the Great Singularity.

AI is not adaptable. AI learns from repetition. Present your AI with a unique situation, and it will not know what to do with it. Imagine you meet a manager who is only equipped to handle decisions they’ve encountered thousands of times before. That, broadly speaking, is your AI.

AI can't validate decisions. There is a lot of uneasiness around letting algorithms decide on issues that require some amount of empathy and understanding of context. Imagine AI making sensitive decisions like adoptions, judging in a court of law, or organ donation. I expect we will see a wave of regulations to tell us that we can use AI as a decision-making tool, but a human needs to do the final validation.

AI rarely follows the process a professional would. AI sees the results and mimics them. It doesn’t grasp the mechanics of things like eyes or hands. AI sees billions of pictures of hands, but doesn’t know how a hand works. If you take your information from photos alone, human fingers could gyrate like a windmill for all you know. This happens with every task where there’s more to the process than meets the eye.

Professional Reconversion at Breakneck Speeds

Each professional so far agreed AI doesn’t catch the essence of their work. An experienced artist knows what’s missing from a drawing. A writer knows there should be more depth to the characters in a book. A chef understands there’s more to recipes than just mixing proteins with greens and adding a dressing.

To a client, though, AI’s best work might be good enough, especially since it comes with a cheaper price tag.

How do we reconvert our jobs?

From Creative to Prompt Engineer. Marketers, Content Writers, Designers

I’ve encountered clients who had a hard time describing their needs. On the first encounter, clients usually (1) can’t tell you precisely what they want, (2) leave out a lot of relevant information, and (3) don’t know much about the process and strategy they require. A client’s first question is “Where do we start?” They are going to have a hard time completely cutting off the middle person.

The market is going to convert some of the creative employees to prompt engineers.

The entry-level is deceivingly low for prompt engineering. There are two UI elements on most popular AI platforms: a text field and a send button. Open the latest Adobe Photoshop version. You will need an AI-generated hand with 50+ fingers to count all the buttons on every side of your screen.

Industries are still trying to figure out how much skill prompt engineers require. AI does a better job than some juniors, at better speeds than senior employees. It’s hard to tell where job automation will happen. Will AI replace the interns who do the repetitive tasks at the office? The employees who make the more sophisticated, high-paying jobs? Is talent becoming overpriced?

We will build an entire pipeline around AI. Who is better at feeding it prompts, and who should validate the results generated?

The professional reconversion to prompt engineers. The best prompt engineers:

• Perfectly understand the process and strategy their work requires;
• Identify the needs that clients can’t express on their own;
• Recognize a good result from a bad one and can improve based on it;
• Have done their research on what the 12 best tools for their job are;

My personal bet: Whether you’re a junior creative or a senior creative, don’t panic, don’t get depressed. The market still needs you. Keep doing your best job and get yourself a good set of tools for the next stage of your career.

There is a lot of discussion going on about the ownership of art. That might prompt a new set of jobs to appear. We are not yet at the Great Singularity. AIs still need to be fed inspiration and be geared towards new trends. Here’s your new mantra: “Whenever we invent a better mouse trap, nature comes with a smarter rat.”

From Data Analysts to Decision-Makers

Imagine you’re in the Olympics. The rules have changed overnight. You can now take as many steroids as you want. As long as you jump really high, nobody cares.

That’s what’s happening with data analysis. It’s a race of AI tools to get the best data crunch. Whoever gets the most efficient palette of machine-learned tools will probably get the gold. With many grinding tasks, data analysts will enjoy the new level of job automation AI tools bring.

We’re living through an experimental time. Programmers managed to build racist AIs, stupid AIs, lying AIs, married AIs, at this point, it’s worse than Tinder. A great AI has to have an outstanding dataset behind it. You’re going to kiss a lot of frogs before finding the good ones.

The professional reconversion of data analysts. The best data analysts:

• Have great tools backed by good, healthy, reliable data.
• They double-check and validate results.
• They can put the data into context.
• They can explain the data to the client and tell everyone in the team how to use it.

My personal bet: When everybody is special, nobody is. With AIs in the race, everybody just became faster, meaner, better. The industry raised the bar. Yes, Coca-Cola will have excellent data analysis, but guess what? So will Pepsi.

There’s a question of how sophisticated the tools become. With writing and drawing, the entry level is fairly low. I suspect AI-based data analysis will come in entry-level, mid-level, and holly-cow-what-is-that-chart level.

Will AIs Be Coming for Programmers Next?

Most of us wouldn’t be able to validate that code. For as much as a TV anchor knows, the generated code could be a button animation, or it could open a hatch to a pool of sharks underneath the floor. When you get outside of your own field, you can no longer validate the work of the AI. You can sire the simple projects, the mom-and-pop website, the light switch app, and the Java animation. I don’t expect BP will use GitHub’s Copilot to program the schedule of their oil rigs.

The sophistication of the job is not apparent when you tell ChatGPT to invent a cocktail recipe for your party. Every professional will tell you generative AIs miss the mark on the subtleties of the job. With programming, this is self-evident. Half of us wouldn’t even know where to stick the generated code.

Still, safety is an illusion. If you know precisely what you want to do and can articulate it in words, a generative AI can deliver a code, document it and explain the process. Some 40% of programming is the grind of repetitive tasks. You know what you want to do, but you need to spell it out, line by line. One elusive mistake will make your code summon demons. A generative AI writes that same code for you fairly well, and part of your job becomes a question of copy-paste. To call it job automation is putting it lightly.

Rules apply here just like everywhere else. If your work gets accelerated by 40%, 40% of the jobs inside a company are now redundant. In a mammoth company, the manager who cuts 40% of salaries spending gets a helluva Christmas bonus.

Some professional reconversion for programmers. The best programmers:

• Will work with machine learning and probably earn four times as much as they’re making now;
• Will discover there’s a great market for code that identifies deep fakes and homework done by ChatGPT.

My personal bet: It is realistic that the software-producing market will simply expand. Clients always felt they could write a fairly good essay or blog article. They rarely assume they could make their own billion-dollar app on their own. Every single one of us started learning how to write in first grade. We started learning how to write code in high school. There is naturally a lot more gatekeeping around it.

Other Ideas for Professional Reconversion

Become a teacher. Here’s the dystopia. Children learn from AI teachers, and AI does their homework. Isn’t that a neatly tied loop? Take humans out of the equation, and you have a supervised machine-learning system.

Teaching naturally stupid humans requires empathy, good communication, initiating conversations, and setting up the learning journey. Whatever your skill is, consider teaching it. Whether we integrate AI into the process or not, teaching human beings will always require humans in charge.

Get an institutional job. Government institutions are slow to change. Because they are working with high stakes, institutions can’t afford to be early adopters.

Your government is not (and should not be) a risk-taker. They have to wait and see if the sharks eat the early adopters, the followers, and the late bloomers before they dive in. You wouldn’t want your government going Vegas-style on your pension fund, even if it’s a high payout slot machine. Consider hunting for a safe institutional job until you make sense of the AI impact.

Make your dream project on a small team. Your boss can make their big dream projects without you? Fine, make your own big dream projects. With blackjack. And friendly people. The market just demonstrated to you that projects don’t need a team that large and a budget that big. Do with that information what you will.

In Closing

The best results still come from humans and machines working together. It’s the same in every single industry. You can’t hand-craft a car, but you can’t build a zero-humans factory and wish it luck in the future, either.

AI is accelerating most white-collar jobs. It doesn’t completely replace neither your star employee, nor your intern. It needs the professional to dictate a process, oversee the work, and validate the results.

Creators know very well generative AIs will not replace them tomorrow. Our concerns are for the next six months.

My bet: Every logarithmic curve looks like an exponential line at first. We expect AI development to rise sharply. We are betting horse and carriage it can’t flatten. In reality, issues like the number of fingers on a hand might be trickier to fix, and these are only the obvious ones.

I believe AI’s journey is more like Moore’s law and will hit a ceiling before it reaches the great singularity and enslaves all humans.

AI tools will eliminate a lot of the gatekeeping. They make small businesses less reliant on creators, and they raise the bar on quality. They will speed up the market. It would be fantastic if we could absorb the growth: generate more business, become faster, and do fewer grinding tasks. The next generation will be richer and more comfortable for it.

Of course, this doesn’t apply to our generation. We’re toast. We’re like a squirrel at a dog convention.