Looking Beyond the Headlines: Thinking Critically About AI News

Aug 16

Reading AI in the News

I am not a scientist, and I do not have a team of researchers or access to deep datasets. Like most people, I learn about artificial intelligence (AI) through the news. Every day, headlines talk about breakthroughs, risks, and massive changes in jobs or society.

Here is the challenge, reading the news is not enough. We need to learn to go a little deeper and reason through what we read.

Too often, stories about AI are written to grab attention. “AI will replace millions of jobs” or “AI will create a golden age of productivity.” Both may contain elements of truth, but rarely do they show the full picture. The data behind those claims, the assumptions made by researchers, and the possible biases in the reporting are usually left out.

A Call to Think Critically

AI is expanding fast, and it is already changing industries and jobs. Instead of being passive readers, we can train ourselves to think more critically about what we see. By questioning, connecting, and reasoning, we get closer to reality, and that is where better decisions come from.

That does not mean we should stop reading AI news. It means we should read differently, ask where the numbers come from, check if other research supports the claim, notice who benefits from the message, and compare multiple sources instead of relying on one.

An Example from Today

For example, today I read a statement that said,

“AI is projected to create 12 million net new jobs in 2025 via automation and upskilling.”

A lot of news like this gets shared and liked quickly. It sounds positive, hopeful, and convincing, and many of us tend to believe it right away. I am not saying these statements cannot be true, but they do require analysis.

The first step I took was to look for another source to see how they were projecting the numbers. I found this in the World Economic Forum’s (WEF) official reports:

In their 2020 Future of Jobs Report, WEF projected a net gain of 12 million jobs by 2025 (97 million created versus 85 million displaced).
In their 2025 Report, which looks ahead to 2030, the outlook changes, showing 170 million new jobs versus 92 million displaced, for a net gain of 78 million jobs.

This strongly suggests the “12 million” figure is not a new projection, it refers back to WEF’s earlier 2020 report.

At that point, I asked myself a simple but important question:

If AI is projected to create 78 million new positions,

How many regular jobs are being taken away in the process?

Difference between “jobs created” and “jobs displaced”

The difference between “jobs created” and “jobs displaced” is what really matters, yet this detail is often left out of the headlines.

Based on research, the numbers show a significant and growing impact on jobs, and many of them will be displaced. Millions of occupational transitions are ahead, and even when risks are not presented as outright job losses, the scale of skill transformation and the balance between positions created and roles displaced remain a major caveat in how real the numbers and scenarios are being presented and interpreted.

When I looked across different studies, I noticed that each source tells the story a little differently, but together they paint a picture of both opportunity and disruption.

Goldman Sachs sees a baseline where 6–7% of U.S. jobs could be displaced by AI under typical adoption, with scenarios ranging from 3% to 14%. They note the peak unemployment effect could add about 0.5 percentage pointsduring the transition, meaning the shift is noticeable but may level out as workers find new roles.
McKinsey talks less about job counts and more about tasks and skills. Their research suggests that by 2030, up to ~30% of hours worked could be automated in the U.S. and EU. That translates into millions of occupational transitions, with about 12 million in the U.S. alone by 2030. In other words, the work doesn’t disappear overnight, but many people will need to shift what they do.
The World Economic Forum frames the issue in terms of balance. In 2020, they projected a net gain of 12 million jobs by 2025 (97 million created vs. 85 million displaced). Their 2025 report updates that to a longer horizon, with 170 million jobs created and 92 million displaced by 2030—a net gain of 78 million. Yet they also stress that this disruption will touch about 22% of current roles, which is not a small number.
The International Labour Organization emphasizes transformation rather than outright loss. They estimate that one in four jobs is at risk of being reshaped by Generative AI, mostly through a changing mix of tasks rather than positions vanishing completely. Still, “reshaping” can mean a very different reality for workers.
The OECD brings another angle, pointing out that across its member economies, about 28% of jobs are in occupations at the highest risk of automation. They remind us that risk does not equal certainty, but it highlights how exposed a significant share of workers already are.
PwC’s AI Jobs Barometer shows the upside for those who adapt. They find that AI-exposed sectors are growing about 3x faster in revenue per employee, and workers with AI skills earn a 56% wage premium. It is a reminder that while some roles may fade, others can thrive if people and businesses are able to make the transition.

The amount of jobs at risk will likely continue to grow, in the past, automation was limited to rule based processes where computers and robots could only follow pre defined instructions, today, advances in artificial intelligence, particularly in machine learning, have broadened the scope. Algorithms are now able to make decisions without pre specified rules, process unstructured environments and data, and perform tasks that were once considered non routine and uniquely human.

This means AI is no longer just replacing repetitive work, but also increasingly capable of handling complex, adaptive, and judgment based activities, as a result, the traditional idea of keeping a “human in the loop” is shrinking, and even when human oversight is still required, the number of people needed can be reduced by 70 to 80%, amplifying the disruptive potential on employment.

References

Goldman Sachs (2025): baseline view that 6–7% of U.S. jobs could be displaced by AI under typical adoption, with a 3–14% scenario range; peak unemployment impact around +0.5 pp as workers transition. Goldman Sachs+1

McKinsey (2025 lens, latest quant still 2023–24): the big shift is task/skill transformation: up to ~30% of hoursautomated by 2030 (US/EU), implying millions of occupational transitions (e.g., ~12M in the U.S. by 2030). McKinsey & Company+1
World Economic Forum (2025): by 2030, 170M jobs created vs 92M displaced globally → net +78M; disruption equals ~22% of current jobs. World Economic Forum+2World Economic Forum+2
ILO (2025): about one in four jobs is at risk of being transformed by GenAI (task mix changes rather than outright loss). International Labour Organization+1
OECD (2025 context): across OECD economies, occupations at highest automation risk ≈ 28% of jobs (risk ≠ certain loss). OECD
PwC AI Jobs Barometer (2025): AI-exposed sectors are seeing 3x faster revenue-per-employee growth and a 56% wage premium for AI skills—evidence that many roles are augmenting rather than disappearing. PwC+2PwC+2

Mikha González