McKinsey published this 60-page GenAI report here which dissects how jobs, skills, and workflow are changing. This gets away from the “oh the sky is falling” generalities and tries to answer the question, “what should professionals actually do next.”
We are not luddites
We’ve seen an explosion of productivity because of the compute, storage, broadband explosion. We believe in Schumpeter’s creative destruction. We are all fans of technology:
- Do you remember when you “Googled with your feet, going to the library?” It’s over-rated.
- My home broadband is 600Mb for about $70 a month. 10 years ago = not available.
- My iPhone camera has 100x picture quality than the digital cameras when I was in MBA
GenAI is here
A few years ago, we might question and gripe about the role of AI in society. In early 2026, this is non-conversation. Anyone in professional services – attorneys, consultants, bankers, professors – have not gone a day in 2025 without the topic of AI coming up with customers, suppliers, and students.
For any professional services person (you), GenAI is a part of our workflow. Fait accompli.
The investment hurricane is incredible. Estimates vary, but it’s quickly going from $250 billion in investment to $1,500 billion (1.5T) to $2,500 billion (2.5T) to something higher. Can’t count that high.
Demand for AI fluency—the ability to use and manage AI tools—has grown sevenfold in two years, faster than for any other skill in US job postings – McKinsey
AI = Alien intelligence
Yuval Noah Harari called AI, alien intelligence, and I get it. It can reason and do things, we thought only humans could do. Scarily, GenAI is the worst it will ever be = today. Unlike other technologies, this thing gets better by being used. . . feedback, data generation, iteration. Even synthetic data (making up it’s own training data).
Basically, GenAI is a hard-working grinder.
work that could be automated: 57%
Not to be alarmist, but McKinsey noted that GenAI (both agents & robotics) could – in theory – automate about 57 percent of current US work hours. Holy $hite. That’s alot.
Basically, non-physical work is 2/3 of US work hours and roughly 2/3 of that (2/3 x 2/3 = 44%) could be heavily automated. Think about all the research, administration, emails, calls, confirmations, invoices, faxes, excel crosswalks. . .all the keyboard based stuff we white collar people do daily. It’s a LOT.
Okay that’s 44%, what’s the rest?
Physical activities are tougher (how ironic is that?) Sure, robots have gotten better (watch Boston Dynamics on YouTube here), but it has a LONG way to go. They just don’t have the motor skills, dexterity, situation awareness, or even the battery life to really make it work. So robots get 13%.
44% agentic + 13% robotic = 57% theoretical automation.
Adoption takes time (thank goodness)
We all know that adoption takes time. Different industries (and the companies within them) adopt differently. Surely, some AI-native new-entrants will go-go-go. While the vast majority (yep, the companies AFTER the chasm) will take their time, wait-and-see, and make incremental changes only.
Electricity took more than 30 years to spread, and industrial robotics followed a similar multi-decade path. As recently as 2023, only about one in five companies ran most of their applications in the cloud, despite the technology being widely available since the mid-2000s. – McKinsey, 2026
What? only 1 in 5 companies run most of their applications in the cloud? So there are a lot of luddites.
people + Agents + Robots
Average has no meaning. “AI is coming for jobs” is a gross generalization. Which jobs? When? Why?
McKinsey did the hard work of analyzing 800 occupations and breaking them into physical (can robots do this job) and non-physical (can agents do this job)? From this fascinating approach, they came up with 8 segments of work, with varying degrees of “uh, oh”.
- People-centric: nurses, psychologists, fire fighters (less change)
- People-agent: sales reps, secondary education, HR specialists
- Agent-centric: Accountants, lawyers, software developers (more change)
- People-robot: Insulation, tile-installers, drywallers
- Robot-centric: Pick/pack logistics, welders (more change)
- People-agent-robot: receptionist, medical assistant, correctional officer
- Agent-robots: machine setters
The more judgment, trust, and human context is needed, the slower the disruption.
The big takeaway is that #1 people-centric were the “safest” from disruption, while #3 agent-centric and #5 robot-centric were probably the most likely to face more immediate turbulence.
GenAI adjacent skills: 600 new ones
McKinsey does wonky analysis (don’t we love this?) They looked at 6,800 skills often cited in 11 million+ job postings and discovered that since 2023, there are 600+ new skills that are in demand. Two thoughts, a) that’s a 10% increase in new types of skills b) they are largely GenAI related.
McKinsey calls these GenAI-adjacent. The point IMHO seems to be that roles are mutating quickly, and the whole idea of one-time “re-skilling” is simplistic. We need highly transferrable skills that we can apply, learn, morph, readapt, and use. Basically, adaptability is better than specialization “when the ground is moving.” We need to think of skills like LEGO bricks. How they are stacked, restacked, keep changing.
Employers are increasingly seeking more AI-adjacent capabilities such as process optimization, quality assurance, and teaching—skills employed to redesign work with AI, supervise and verify AI systems, or train people to use them. – McKinsey
Which skills are in less demand? more demand?
The McKinsey report is 60 pages and worth your time. Even if it is only 1/3 correct, this is worth paying attention to. On page 27 (exhibit 9), it shows skills that are mentioned LESS and mentioned more. As the highway signs sometimes say, “stay right.”
Value migrates upward
It makes sense that all professionals will gravitate to more value-added work. Instead analyzing data, we will be interpreting findings. Instead of forecasting demand, we will be creating scenario plans. The machine will do more of the mis-en-place (washing, cutting of vegetables), which will give us more time to hone recipes, do quality control, and engage with our (metaphorical) customers. When we think of this in terms of a maturity model, we will be doing level 4 an 5 work, not level 2 work.
Thought experiment
Time to eat my own dog-food. As a teacher and trainer, what work is less likely to go to the machine?
I am guessing, but here goes:
- Curating of content, determine which videos are relevant, useful, not repetitive, engaging, and authoritative
- Facilitating a live discussion, creating a safe place for constructive tension, debate, and synthesis
- Qualifying a potential client project for scope, budget, and fit
- Following up with a student because they didn’t understand XYZ during a class discussion
- Holding an office hour with a student who is weighing the pro/con of a specific internship
So What?
1) Skim through the McKinsey report yourself here
2) Download the report, upload it into ChatGPT, and then ask it to help segregate YOUR specific work so that:
- You spend less time on keyboard work
- You spend more time on judgment work
- You invest in transferrable skills
- You help with decisions, not the research