This weekly update brings mixed news for workers dealing with AI agents in their jobs.

Major companies are pushing hard for employees to learn AI skills. Citi bank started training all 175,000 workers on how to write better prompts for AI tools. The bank says employees have already used AI tools over 6.5 million times this year. Other big companies like Accenture are even firing workers who can't learn to use AI.

But many employees are struggling with AI-generated content that actually makes work harder. A new study from Stanford University found that 40% of US workers dealt with "workslop" - AI content that looks good but doesn't really help. This low-quality AI work costs companies about $186 per employee each month in lost time. Workers have to spend extra hours checking facts and rewriting what AI produced.

The good news is that AI isn't stealing jobs as fast as people feared. Research from Yale University shows that since ChatGPT launched in late 2022, the job market has stayed pretty stable. The mix of different types of jobs hasn't changed much more than during past technology changes.

At Workday Rising 2025, company leaders talked about how AI should help workers, not replace them. They introduced "Everyday AI" to help employees feel more comfortable using these tools. The company found that workers who try new AI projects are 42% more likely to get promoted and 30% less likely to quit.

However, many employees still don't trust AI tools. A British study found that 64% of workers think AI agents are unreliable. More than half said AI tools give wrong answers but act too confident about them. Some workers are even hiding their AI use because they worry it might hurt their reputation.