A.I. will probably most likely lead to the end of the world, but in the meantime, there’ll be great companies.
Whatever the future may hold for the field of artificial intelligence and humanity, it’s inevitably going to be disruptive. Yes, people are definitely going to lose their jobs to AI. Yes, entire industries will probably be upended and replaced with something we can only vaguely imagine today. And sure, maybe a rogue AI will wipe us all out, but, as Sam Altman says, there’s definitely room for some cool companies to emerge in the meantime.
Cool companies can lead to cool jobs. And the thing about disruption is that it doesn’t lead entirely to losers… some winners come out of the fray, too.
Today, that means that some folks who are forging new and unusual career paths in artificial intelligence are very likely going to land in the driver’s seat for the next generation of AI.
A Quick Look at the Eight Lane Highway into AI Careers Before Exploring the Backroads
With a field as new as artificial intelligence, it’s fair to ask what exactly constitutes a conventional career path. After all, very few people have been working in the industry long enough to go from college to retirement. The state-of-the-art changes swiftly and it’s really only been the past few years that anything like a stable demand has existed for AI professionals.
But we’re already seeing some well-worn grooves come out of the AI explosion in jobs like:
- AI Engineer
- Computer Vision Engineer
- AI Programmer/Software Developer
- AI Architect
For the most part, despite the complexity of the subject matter, these reflect typical job roles that have long existed in the IT industry. They also stay very much in the typical lanes of IT career progression:
- Undergraduate and then graduate studies in math, computation, and coding
- Professional certification in tools or techniques commonly used in the industry
- Challenging coding and logic interviews with mainstream or startup tech companies
- A life of open-plan offices, ping-pong tables in break rooms, and free snacks and drinks in the company cafeteria
If AI is something entirely new, though, what about the entirely new and visionary kinds of jobs that it may create? Just as interesting, what paths will people take to land those positions?
The Least Conventional Paths to AI Work Will Start Outside the Core of the AI World
Most of the strangest jobs in AI probably won’t come near the core of the profession. As it turns out, architecting and crafting code that build the brains of AI resembles most other kinds of data science and software development work… a lot of calculation, coding, building, and testing.
Instead, the funkiest new AI jobs will come at the periphery, where pure scientific developments meet applied uses in the business world.
Out here, where the sometimes-strange logic and behavior of the current crop of AI requires some massaging to meet up with customer expectations and business needs, jobs like AI prompt engineers and observability engineers will grease the gears.
An Unconventional Career Starts from Unconventional Preparations
Just as interesting as these new positions themselves are the unconventional ways that people come to them. There are no degrees for prompt engineering or in machine vocational training; a philosophy degree might be considered pretty useful to any ethicist, but they certainly aren’t common in the computer industry.
So not only are the positions themselves unusual, but the paths to land them are equally outside the mainstream of AI education. Poets and linguists can have as good a shot at these jobs as dyed-in-the-wool computer scientists. Maybe the best way in is a unique combination of training and experience that bridges the gaps between the hard sciences and the new psychology of the machines.
Prompt Engineer: The AI Whisperers That All Large Language Models Will Need
Prompt engineers are already getting a lot of press. Anyone who has ever asked ChatGPT for anything particularly specific may have noticed that it isn’t always as quick on the uptake as another person. This isn’t surprising if you think about it. Even conversations between two people can spark misunderstandings or misinterpretation.
While people can use context and the great depths of their personal experience to head off most of these misunderstandings, large language models don’t have anything like that to fall back on. What they know is their training data and a very limited range of conversational memory.
Prompt engineers bank on their knowledge of that training data and the scope of that conversation to craft statements and queries in the way most likely to produce the results they are looking for.
How to Start Down the Path to AI Prompt Engineering
The number one requirement for a prompt engineer is that they have a way with words. That can come from anything from a degree in English to a natural gift of gab. You’ll spend a lot of time chatting with machines and restating similar queries in different ways. That means you had better bring a strong vocabulary to the table and have wicked thesaurus skills on tap.
Of course, there are still important AI concepts that are useful to understand. Most importantly, a grasp of logic and reasoning patterns exhibited by machine learning algorithms is pretty important. And, of course, you have to get hired—that means networking within the AI industry.
This combination of skills may mean earning a Bachelor of Arts in AI is still a good bet, even if it’s not entirely required. You’ll meet the people you need to meet and master the essential concepts of the field.
AI Ethicist: Determining Right and Wrong in the New World of Machine Intelligence
Big concerns about social justice, equity, and the effects on society from a sudden influx of dubious AI logic have many people asking if AI is a good idea in the first place. Even among those who believe in the technology, there’s an impulse to pump the brakes right now and try to figure out what it all means. Almost 34,000 experts and developers signed an open letter in March 2023 requesting a pause for consideration on all AI experiments.
That pause in AI development never happened. In fact, it accelerated.
Enter the AI ethicist. Combining a deep understanding of what AI is and what it does with insights into human behavior, social norms, and the expectations we share for fairness and reason, ethics professionals offer input on how AI is being built and how it can best be used.
While no one is slowing their development efforts, more and more companies are hiring professionals to make sure they are done in a responsible and sustainable way.
How to Get Started as an AI Ethicist
Determining what’s desirable from among what is possible does still require strong technical understanding in this job. It does not, however, need to be specific to artificial intelligence. A degree in ethics or philosophy might be the best grounding for these roles. Even law degrees can help build your understanding of norms and the reasoning skills to apply to questions of morality.
Following up that essential education in logic, reasoning, and human society with some time spent working in the information technology industry, or at least studying computer science, may be all you need to launch into this new field.
Of course, experience is the real differentiator in any kind of ethics job. No one wants to hire a fresh-faced youth with a shiny new bachelor’s degree to weigh in on matters of life and death.
So expect to put in some time in social sciences, the law, or academia before getting your hands dirty as a professional AI ethicist.
AI Trainer: Teaching Machines the Right Values
You create your brain from the input you get.
Machine learning implies teaching.
Just like people, machine minds are constructed from the information they are primed with. In many cases, the ML behind artificial intelligence projects uses unsupervised learning, which allows it to crank away against training data on its own. It can find patterns and infer meaning from those efforts with some astonishing results.
But it turns out that in most cases, having actual human oversight and correction in the training process produces even better results.
One of these methods, supervised learning, is a straightforward process of labeling training data and feeding it to the machine with relationships already baked in. Somewhat famously, it’s a process that anyone can do, as illustrated by the original development of ImageNet using random Mechanical Turk workers.
Other types of training, however, require experts. Methods like active learning, where algorithms actively ask questions from human users, or apprenticeship learning, where the machine observes the tasks performed by a human operator, or the popular reinforcement learning from human feedback, where a person offers corrections to algorithmic guesses, all require trainers with advanced knowledge.
While some of that knowledge is of AI systems and how the underlying algorithms are built and perform, much of it is about the domain where the training is happening.
No one is going to put a pasty, basement-dwelling machine learning engineer behind the wheel of a Formula 1 car to teach an AI how to drive.
So paths into this area of the AI industry will typically involve people coming from outside that industry. They’ll need special domain knowledge as well as insight into AI systems.
How to Start on the Path to AI Training Jobs
Many AI trainers are getting their extra expertise in machine learning systems on the job. For example, companies working on AI systems in legal services need fully trained and experienced lawyers and paralegals to evaluate and guide algorithms.
It’s far easier for a lawyer to go back to earn a certificate in artificial intelligence to pick up some rudimentary exposure to math, statistics, and coding than it is for a programmer to go withstand the pressure cooker of law school. In fact, with the right generative front-end on AI systems, much of the work of training can happen in plain English.
Unlike some of the other new jobs in AI, working in AI training will still require a highly specialized education and a tough path to a professional position. But it will be an unconventional path from the AI engineering perspective.
Of course, conventional or unconventional, what matters most is that you find a stable position in a field that you love. AI may or may not be your one true love, but as the clear technology focus of the future, it’s going to be one of the most stable industries to join.