Not all upskilling investments produce the same protection against AI disruption. Generic advice to learn new technology or improve soft skills misses the question of which specific skills have the highest protection value in a world where AI systems are advancing rapidly. The four categories with the strongest evidence for durable value are: prompt engineering applied to your specific domain (knowing how to use AI tools well in your field), AI output evaluation (the ability to judge whether an AI-generated result is correct and appropriate), deep domain expertise that provides the context AI outputs lack, and interpersonal judgment in high-stakes situations where trust and human accountability matter. These are not equally accessible to everyone starting from scratch, which is why the right approach is to identify which of these you already have partial advantage in and build from there.
What the research says
92M
jobs displaced by 2030
WEF Future of Jobs 2025
170M
new roles created by 2030
WEF Future of Jobs 2025
41%
of employers plan AI-driven headcount reductions
WEF 2025
55K
job cuts explicitly attributed to AI in 2024
Challenger, Gray and Christmas
Prompt engineering is the skill of structuring inputs to AI systems to produce reliable, high-quality outputs. It is worth learning as a baseline skill in 2026, but it is not a career in itself for most workers. Its value is highest when combined with domain expertise: a lawyer who can prompt AI legal tools effectively has a significant workflow advantage over one who cannot.
AI output generation is increasingly accessible to everyone. What remains scarce is the ability to accurately assess whether an AI-generated output is correct, complete, appropriate, and safe to use in a professional context. This evaluation skill requires deep domain knowledge and cannot itself be easily automated. It is the skill that justifies a human in the loop.
Interpersonal judgment includes reading a room, navigating political dynamics, building trust with specific individuals over time, and making calls that account for unspoken context. These functions require social embodiment and relationship history that AI systems do not have. Workers who are visibly strong in interpersonal judgment tend to be assigned to higher-stakes human interactions, which are among the last tasks to be automated.
For most non-technical professionals, basic coding is a lower-ROI investment than AI literacy and domain deepening. AI tools can generate substantial amounts of functional code, which has reduced the competitive advantage of basic coding skills. Software engineering at a senior architecture and system design level remains highly valuable, but coding as a standalone skill is more commoditised than it was five years ago.
Measure upskilling return by tracking two things: whether you can take on higher-complexity tasks than before, and whether you are being assigned more judgment-dependent work rather than execution work. Pay rate and title changes follow those leading indicators. If neither is shifting after 12 months of upskilling effort, the content or direction of the investment needs to change.
Free Assessment
Free Fossil Score assessment. No account required.
Get My Fossil Score