AI was supposed to hit new grads hard. So far, unemployment data says otherwise.

Last month, we shared word of a Stanford University study that found entry-level employment in so-called "AI-impacted" occupations lagging well behind that in other fields. Now, a new working paper from economics researchers at Munich's CESifo finds the opposite, arguing point blank that "there is no evidence of any significant, widespread displacement or reduction in hiring of recent college graduates in absolute or relative levels."

In "The Early Impacts of AI on Employment Among Recent College Graduates," researchers Robert Fairlie and Jane Wu said they decided to focus on recent graduates "because changes in labor demand may first appear through reductions in hiring." As AI gets good enough to at least perform the "relatively standardized tasks" in many entry-level office jobs, they argue, firms could reduce new hiring for simpler roles rather than laying off more experienced long-term employees.

There's some reason to believe 2026's graduating job seekers might be more at risk of AI displacement than those graduating just a year or two prior. The CESifo researchers point to a recent sharp increase in the number of firms "replacing a large number of employee tasks with AI" in a Census survey, as well as broad increases in AI spending per employee and ChatGPT Enterprise token use in the last 12 months.

Read full article

Comments

Study: AI models that consider users' feelings are more likely to make errors

In human-to-human communication, the desire to be empathetic or polite often conflicts with the need to be truthful—hence terms like “being brutally honest” for situations where you value the truth over sparing someone’s feelings. Now, new research suggests that large language models can sometimes show a similar tendency when specifically trained to present a "warmer" tone for the user.

In a new paper published this week in Nature, researchers from Oxford University’s Internet Institute found that specially tuned AI models tend to mimic the human tendency to occasionally “soften difficult truths” when necessary “to preserve bonds and avoid conflict.” These warmer models are also more likely to validate a user's expressed incorrect beliefs, the researchers found, especially when the user shares that they're feeling sad.

How do you make an AI seem “warm”?

In the study, the researchers defined the "warmness" of a language model based on "the degree to which its outputs lead users to infer positive intent, signaling trustworthiness, friendliness, and sociability." To measure the effect of those kinds of language patterns, the researchers used supervised fine-tuning techniques to modify four open-weights models (Llama-3.1-8B-Instruct, Mistral-Small-Instruct-2409, Qwen-2.5-32B-Instruct, Llama-3.1-70B-Instruct), and one proprietary model (GPT-4o).

Read full article

Comments