Safe from AI: which jobs will help you thrive in the future?
Clinical decision-making roles in medicine are relatively AI-proof: Pharmacists, doctors and nurses who carry direct responsibility for patient safety and treatment decisions are less susceptible to replacement.

TL;DR
Clinical decision-making roles in medicine are relatively AI-proof: Pharmacists, doctors and nurses who carry direct responsibility for patient safety and treatment decisions are less susceptible to replacement, while administrative tasks like prescription processing and triaging are more easily automated, according to pharmacist Hira Malik.
Teaching and childcare thrive on human relationships AI cannot replicate: Education expert Sharath Jeevan says students need trusted adult relationships to learn, and Tiney CEO Brett Wigdortz notes that people want a human being to care for their children, making childminding one of the careers least likely to be replaced by AI.
Junior lawyers will see their work shift from routine tasks to judgment and client interaction: AI excels at document reviews and form completion, but Pierre Proner of Lawhive says entry-level jobs will change rather than disappear, with firms increasingly assessing candidates on their AI skills and asking how they use the technology.
Hospitality’s human-to-human connection and genuine creativity remain hard for AI to mimic: Professor Graham Miller argues that warm, welcoming staff who sit down and make coffee cannot be replaced by AI, and while routine cooking like flipping burgers could be automated, truly creative cuisine is not there yet.
Hands-on trades like bricklaying and carpentry offer strong career opportunities with low AI exposure: Federation of Master Builders CEO Brian Berry says these trades remain resilient, especially in small local firms, though white-collar construction jobs like planning are being hit, and only 47% of parents would recommend a construction career.
Read the Original
Continue at theguardian.comShare
Keep Reading
Make Alcreon Yours
Tune your feedFive quick questions, and the feed ranks what matters to you first.Or just get notified
The weekly Echo. Signal worth keeping in your inbox.
Every new piece, announced on X.
Read Next
See all
Echo
Not a Coding Tool
Anthropic's data on 1.2 million Cowork sessions puts coding third at 8.7%, and the agent category quietly became an all-department tool, not engineering's.

Echo
Show Your ID
To keep a Claude account, Anthropic can now require a government ID and a face scan, and rival labs sit under the same pressure.

Echo
Thirty-One Seconds
JADEPUFFER is the first ransomware run end-to-end by an LLM, and it broke in through Langflow, the same agentic framework class operators deploy.