Living With AI
Artificial intelligence has become part of the daily conversation almost overnight. Some people are excited about it, some are worried about it, and most of us are simply trying to understand what it actually means.
One of the questions I hear most often is a simple one: Which jobs will AI affect first?
If you step back and look at how these systems work, a pattern begins to appear. AI is very good at reading information, summarizing it, answering questions, and following structured rules. That means the first changes are likely to appear in jobs that involve large amounts of digital information and routine decision-making.
Below are a few areas where analysts expect the earliest shifts.
Areas Where AI May Change Work the Most
| Job Area | Approximate Workforce | Why AI Fits |
| Customer service / call centers | ~3 million (U.S.) / ~20 million (global) | answering questions and troubleshooting |
| Administrative support | ~22 million (U.S.) / ~150 million (global) | scheduling, organizing information, preparing documents |
| Bookkeeping & routine accounting | ~2 million (U.S.) / ~12 million (global) | structured financial rules and data |
| Legal research & paralegal work | ~350,000 (U.S.) / ~2–3 million (global) | reviewing contracts and searching case law |
| Marketing content & copywriting | ~1.5 million (U.S.) / ~8–10 million (global) | generating routine written content |
| Translation & transcription | ~70,000 (U.S.) / ~500,000+ (global) | speech recognition and language translation |
One important point is worth keeping in mind.
Technology rarely eliminates an entire profession overnight. What usually happens instead is that the same work can be done by fewer people. One person with powerful tools can sometimes do the work that once required several.
That pattern has appeared in almost every technological shift we’ve experienced, from tractors on farms to computers in offices.
The real question may not be whether AI replaces jobs. It may be how the way we work gradually changes as these tools become part of everyday life.
This series is simply an attempt to watch that process as it unfolds.
A Question Worth Considering
Where do you see AI already beginning to change the work people do around you?
If you have an example, I’d be interested to hear it in the comments below.
Living With AI Series
Next in the series:
The Pattern Most People Miss
will be posted on 3/10/26
One place I’m already seeing this is transcription. A year ago turning an audio recording into text was slow and expensive. Today it can happen in minutes. I’m curious where others are seeing similar shifts.
Hey Rich, probably no surprise but I have opinions on the rise of AI, like you having seen many “job killers” from tech firms emerge over the last 30+ years.
One of the first things to get straight is that we have been living with AI in one form or another for decades. The rules based capabilities on many commonly used PC tools are basic AI, the call handling applications (IVR/IVA) and the rather questionable chat bots on many websites.
What makes this iteration different is what sits behind the AI applications. No longer does AI only use the intelligence from internal processes and analysis along fixed or predefined paths. The advent of LLM’s (large language models), the compute power in data centres, the access to massive broadband speeds and the construction of agentic and generative AI architectures, results in levels and quality of output we could only imagine a few years ago. Most AI we experience is not one thing, it’s a combination of capabilities that can be orchestrated by one “agent” written to do this specific task, combining with research or transactional agents that bring information together. This can then be analysed by an agent designed to check the quality of the response, compare this to other similar questions and responses before deciding what to present back. All of this is happening at lightning fast speeds which enhances the impression that these agents or programs are intelligent enough to replace humans…full stop ( or period as you like to say). The challenge we and AI face is that unless the output is qualified, checked, reviewed in a way that most humans would do, what AI creates can be dangerously wrong as often as it is dazzlingly creative. Responsible AI is a phrase we will begin to hear more often, it’s hugely important that we don’t just accept what is presented. AI can hallucinate and will fill in gaps with unconfirmed information to complete its task, blindly accepting this information is dangerous and damaging in many ways.
Don’t get me wrong, I’m a huge advocate for AI but it has to be used responsibly and applied where it can add the most value, not just thrown at every opportunity because it is flavour of the month!
All the best my friend
Mawgan, that’s a very helpful perspective, especially from someone who has watched several waves of technology come through the workplace.
You’re absolutely right that AI didn’t suddenly appear out of nowhere. As you point out, many of the tools people have been using for years — IVR systems, rule-based automation, even some of the early chat systems — were forms of AI, just far more limited in how they could process information.
What seems different now, as you describe, is the combination of things happening at once: large language models, enormous computing power, access to vast amounts of information, and the ability to orchestrate multiple systems together to complete tasks. The speed and scale of that combination is what makes this moment feel different.
Your point about verification is an important one as well. One thing I’m discovering while working with these tools is that they can be remarkably useful assistants, but they still need human judgment. Used carefully they can accelerate thinking and research, but used carelessly they can certainly lead people in the wrong direction.
I appreciate you adding that perspective my friend.
It seems like many professions are discovering the same thing — AI is speeding up the research and information parts of work first.
I have embraced AI in healthcare. I used to literally spend hours a week researching potential client diagnoses, diagnostic tests, and treatments. AI could give me all that information in minutes. As John Dewey said, “A problem well put is half-solved.” Asking the right question is a lot of it. And, of course, information has to be verified. Harvard School for Public Health offers a course about AI in healthcare. They project a 50% reduction in treatment costs when using AI for diagnoses; 40% projected improvement in health outcomes when using AI for diagnoses; and a $187.9 estimated market for AI in health care in 2030!
Renee, thank you so much for your comment; that’s a great example of the kind of shift I was thinking about when writing this first piece.
Medicine seems like a field where access to large amounts of information — research papers, case histories, diagnostic possibilities, treatment options — has always been part of the work. What you describe is exactly where these tools seem to help most: quickly gathering and organizing information that once took hours of searching.
I also appreciate your point about verification. One thing I’m learning while working with these systems is that they can be very helpful assistants, but they still require human judgment and experience to interpret the results.
It will be interesting to see how this evolves in healthcare. If doctors can spend less time searching for information and more time with patients, that could be a very positive change.
I’ve actually been thinking about exploring the use of AI in healthcare in a future essay. If you’d be willing, I might run that piece by you before publishing and get your thoughts.