AI in Business
How accurate are AI patent counts? A new tool suggests the standard measure misses most of them
A new classifier for AI patents finds the standard USPTO tool misses most real AI inventions. Applied to U.S. and Chinese filings, it reveals convergence in AI output but sharply different institutional and geographic patterns.
Can an algorithm judge a future leader? A large-scale test of AI scoring in hiring simulations
A new study tests whether an AI trained on 33 million words of leadership assessments can match human experts at scoring candidates and predicting career progression.
Can ChatGPT beat the S&P 500? Eight months of daily picks suggest no
Tracking eight months of daily stock picks from ChatGPT, Claude, Gemini, and Grok, researchers find AI portfolios are concentrated in large tech names, tilted toward momentum, and deliver no abnormal returns after accounting for those tilts.
When your job feels scriptable: How routine work and AI anxiety drain employee energy
New research on 550 frontline workers finds that highly scripted jobs drain vigor and learning, and AI-related job anxiety disrupts the knowledge sharing that normally helps employees thrive.
Can AI read the room? How news sentiment signals which stocks will bounce back after a crash
A new study of 2.7 million news articles finds that AI-generated sentiment scores can help distinguish temporary stock crashes from lasting declines, though the signal is getting faster.
How AI is rewriting the marketer’s playbook, according to a wide-ranging literature review
A literature review of more than 200 publications maps how AI is being woven into marketing, from programmatic advertising to churn prediction, and where adoption still stalls.
Can GPT-4 pick stocks? A new AI framework reports market-beating returns on the S&P 100
Researchers built an AI framework around GPT-4 that reads news, earnings reports, and macro data to pick stocks. On the S&P 100, it beat the index by 10 to 30 percentage points over 15 months.
When two heads aren’t better than one: What research reveals about human-AI teamwork in marketing
A meta-analysis of 132 experiments finds that adding AI to marketing workflows reliably improves human performance, but human-AI teams rarely outperform capable AI systems working alone.