Johannes Wachs
Researching social computing, crowds, and networks at Corvinus University of Budapest and HUN-REN CERS.
More at: johanneswachs.com
- In a new Science study, researchers train a neural classifier to spot #AI-generated Python functions in over 30 million GitHub commits by 160,097 software developers, tracking how fast, and where, these tools take hold. scim.ag/4aeUdAV
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- Bis Ende 2024 wurden schätzungsweise rund ein Drittel aller neu geschriebenen Python-Funktionen in den USA mit KI-Unterstützung erstellt. Weniger erfahrene Entwickler nutzen AI-Tools öfter. Nur erfahrene Entwickler:innen erzielen erkennbare Produktivitätsgewinne. www.science.org/doi/10.1126/...
- AI writes about one-third of new software code in the US (29%), with large global differences (e.g., Germany 23%, Russia 15%), shows a new CSH study published in @science.org. This brings productivity gains, but so far seasoned software engineers are the only ones who benefit. shorturl.at/LPV6C
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- In a recent piece for the @wsj.com commentator @greg_ip cited a 2024 PNAS Nexus study by @johanneswachs.bsky.social + coauthors @maria-drc.bsky.social & N.Laurentsyeva showing that LLM can be a potential substitute for human-generated data & knowledge 📚 www.wsj.com/tech/ai/will...
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- “Our conservative model finds that going from 0→30 % AI share (US 2020-24) predicts 2.4 % increase in commits. Using task & wage data on occupations, this implies genAI creates $9-14 bill/year in US software alone. Larger estimates of effects from RCTs imply $100 billion.“ arxiv.org/abs/2506.08945
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- 🧵🧪 New paper alert! We studied how firms consume information by analyzing online reading patterns across millions of organizations. Some fascinating patterns emerged... (1/7)
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
