Tech workers spend nights on AI, fearing irrelevance

From Dublin to San Jose and Seattle, tech workers say AI has boosted productivity at work while stealing time after hours. Many are learning tools like Cursor, Claude Code and ChatGPT on weekends—sometimes paying out of pocket and racking up hundreds or even t

By the time Maahir Sharma’s Dublin workday ended, his AI project was already negotiating hotel rooms in the United States.

One night after work, the 24-year-old software engineer watched an AI agent he built place calls and negotiate room rates on his behalf. The project wasn’t part of his job. It was one of many experiments he pursues outside work to stay current in an industry being reshaped by AI.

Sharma says AI has dramatically increased his productivity—helping him complete some tasks in days that once took months. But he also spends about 20 hours a week experimenting with tools after work, including Cursor, a coding assistant he pays for out of pocket.

“I think experimentation with AI is very important,” Sharma said. “If you don’t have hands-on experience, it could be difficult to survive in the industry.”

His experience reflects a broader shift desk workers are describing across tech and beyond: the technology is giving them time back at work, while taking time from evenings and weekends.

A survey by Ernst & Young of more than 1,000 US desk workers across six industries, conducted last year, found that 85% were learning how to use AI outside of work.

For many, the after-hours grind is powered as much by competitiveness as curiosity. Meta and Microsoft have offered multimillion-dollar compensation packages to top AI talent even as both companies have laid off thousands of workers in recent years. Hiring for AI engineers on LinkedIn has surged since 2022. while hiring for many traditional engineering roles has remained flat or declined. according to data shared with Business Insider.

The pressure isn’t only about learning new software—it’s about staying employable as job expectations move faster than training cycles.

Tanvi Pisal felt that speed in her own career arc.

In early 2025. Pisal. then a product designer at an AI healthcare startup in San Jose. began to worry that AI could be coming for her job. She said a company leadership summit underscored how quickly AI was advancing. raising concerns that some UX and product design tasks could eventually be automated.

She started expanding her AI skills and exploring other opportunities. Then, last October, she was laid off. An email accompanying the cuts said they were tied to the company’s rapid adoption of AI.

Now a UX design contractor for a Big Tech company. Pisal spends 10 to 15 hours a week outside work learning about AI. Her routine includes experimenting with tools and attending workshops. She has also spent hundreds of dollars on AI tools and workshops, including subscriptions to ChatGPT and Claude.

“If I don’t spend a few hours over the weekend catching up on updates, experimenting with tools, or reading about what’s new, I start falling behind,” said Pisal, who is 29 and lives in San Jose.

Not everyone agrees the problem is a training gap.

Some workers argue the bigger constraint is time. Even those using AI extensively at work say their day-to-day responsibilities leave limited room to explore the expanding number of AI tools and models. The challenge isn’t only keeping up with what matters now—it’s understanding which tools may matter tomorrow.

Still, the burden doesn’t land the same way on everyone.

Manoj Aggarwal, a lead engineer at a large software company, spends a couple of hours a week outside work experimenting with AI tools and about $60 a month on subscriptions. He said his employer provides access to many of the latest AI tools, letting him develop AI skills on the job.

Much of his reading and experimentation happens after his young daughter falls asleep.

Udit Mehrotra, a head of product at Amazon, spends roughly five to seven hours a week outside work experimenting with AI. He said that last December, he built 10 apps in about a month, working evenings and weekends with Claude Code as his main assistant.

In recent months, he has tried to learn in a more sustainable way.

“I’ve come to think of this less like a sprint and more like a marathon,” said Mehrotra, who is in his 30s and lives in Seattle.

An Amazon spokesperson said in a statement that the company provides employees with AI training and learning resources. including an internal hub that helps workers identify AI tools relevant to their work. The spokesperson said Amazon encourages employees to experiment with AI as part of their day-to-day work.

Even within Amazon, the time math can be brutal.

Abhinav Bohra, a senior applied scientist at Amazon based in Seattle, spends roughly eight to 12 hours a week outside work keeping up with AI. He said he spent about $3,000 over the past year on AI tools, conference fees, and professional memberships.

Much of Bohra’s AI learning happens on evenings and weekends because his workday is consumed by meetings and deliverables. He described the result as a “learning tax” that blurs the line between professional development and personal time.

“Continuous learning has quietly become part of the job, even when it happens outside the job,” said the 32-year-old.

For Bohra, the worry isn’t a single tool replacing him overnight.

“The concern isn’t that one AI tool will replace me overnight,” he said. “The bigger concern is becoming technically stale in a field where the baseline is constantly moving.”

The thread running through these accounts is simple: AI is rewriting how work gets done, and workers say the industry’s speed forces them to rewrite their own lives around it. When time after hours becomes part of the job itself, the line between keeping up and keeping control starts to disappear.

For Sharma, Pisal, Aggarwal, Mehrotra and Bohra, the decision is rarely framed as optional anymore. The tools keep changing. The expectations keep shifting. And in a field where they believe the baseline is always moving, nights and weekends are becoming the cost of staying in the game.

AI tools tech workers Cursor ChatGPT Claude Claude Code productivity layoffs upskilling Amazon Microsoft Meta Ernst & Young survey