73% of tech jobs now require AI skills, up from 15% in 2024

Key Takeaways

• 73% of US tech job postings now require at least one AI skill, up from 15% in January 2024
• Certifications from AWS and Google are emerging as proof of AI proficiency
• Employers want candidates who can audit and manage AI tools using domain expertise, not just defer to them

Nearly three-quarters of tech job postings in the US now list AI skills as a requirement, according to a Dice report analyzing 7 million listings from May 2026. The figure stood at 15% in January 2024. That's not a gradual shift. It's a collapse of the old hiring baseline.

"A lot of these skills are going to just become table stakes," Dice CEO Art Zeile told ZDNet. For tech workers already dealing with a rocky job market, the question isn't whether to learn AI tools. It's how to prove you actually know them.

Why certifications suddenly carry weight

Two years ago, AI certifications barely existed. Now AWS, Google, and others offer credentials for generative AI developers, machine learning engineers, and related roles. Zeile sees them as a direct signal to hiring managers: you went through a training program, passed a test, and can prove it.

Certifications have always been a shortcut for HR departments to filter candidates. The difference now is that the AI cert landscape has matured enough to matter. If you're job hunting in 2026 without at least one credential tied to AI or ML, you're competing against candidates who have them.

Talking through projects beats listing skills

Saying you know Python isn't enough. Zeile points to Columbia University's career guidance: don't list generic duties on a resume, but what you accomplished, how, and why. A project that saved your last company time or money tells a story. Listing "proficient in Python" does not.

That could mean walking into an interview with an agent you've built. It could mean pointing to a workflow you automated. The specificity matters. Hiring managers are drowning in resumes that claim AI fluency. The candidates who get callbacks can show the receipts.

Domain expertise still wins

Dan Hillman, an interview engineer at Karat, runs technical assessments for clients including Google, Goldman Sachs, and Mastercard. He's looking for something specific: how well candidates use their own expertise to audit and manage AI tools, not just defer to them.

“It's not about testing only how well you can work with AI. It's testing how well you work in your domain, augmented by AI.”

— Dan Hillman, Interview Engineer at Karat

His advice: practice problems ahead of the interview, using AI. Come up with your own approach first, then work with the tool, then review the output. That sequence builds the muscle memory interviewers want to see. You're not outsourcing your thinking. You're augmenting it.

Employers want to see your upskilling plan

Michael Morris, global head of platform and talent at Randstad Digital, doesn't mince words: "Job seekers today that don't come in with a real training and upskilling personal plan, I wouldn't consider them."

Online courses help tech professionals stay nimble as new models emerge. But the plan itself is what matters. Can you articulate how you'll keep pace as the tools change every six months? Do you have a strategy for when your specialty becomes vulnerable to displacement?

Morris says candidates need to show they understand how their role might be affected by AI advances and have a plan to adapt. That's a higher bar than "I took a ChatGPT tutorial." It's asking you to think about your career trajectory in a way that acknowledges the ground is shifting.

What interviewers actually test for

Hillman emphasizes process over output. He wants to hear how you gather information up front, write specific prompts, question what the AI gives you, and budget your time. The candidates who struggle are the ones who accept the first response and move on.

That skepticism is the skill. AI tools generate plausible answers whether or not they're correct. The ability to catch errors, ask follow-up questions, and validate outputs against your domain knowledge is what separates a developer who uses AI from one who is used by it.

Frequently Asked Questions

What percentage of tech jobs require AI skills in 2026?

According to Dice's analysis of 7 million US job postings in May 2026, 73% now require at least one AI skill. This is up from 15% in January 2024.

Which AI certifications do employers value most?

AWS and Google certifications for generative AI developers and machine learning engineers are emerging as recognized credentials. The specific cert matters less than having verifiable proof of training.

How do I demonstrate AI skills in a job interview?

Come prepared to discuss specific projects where you used AI tools. Explain your process: how you prompted, how you validated outputs, and what results you achieved. Bringing an agent or automation you built is even better.

Is domain expertise still important in an AI-driven job market?

Yes. Employers want candidates who can audit and manage AI tools using their own expertise, not defer to AI without question. Your specialty combined with AI fluency is more valuable than AI skills alone.

Should I have an upskilling plan for AI?

Hiring managers increasingly expect it. Be prepared to explain how you'll keep pace as tools evolve and how you'll adapt if AI affects your specific role.

Source: Latest news