FEW USE AI DAILY

Plus Zuck Regains Cred (subscribe in the links below)

Get a free 20-page AI explainer: AI FROM ZERO plus these stories and more, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us

Like this? Get AIDAILY, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us

Hardly Anyone Actually Uses Generative AI

A study by the Reuters Institute and Oxford University reveals that while ChatGPT is the most recognized generative AI tool, only a small percentage use it daily. Younger people are more likely to use these tools, but overall regular usage remains low.

Bilingual AI Brain Implant Helps Stroke Survivor Communicate

Scientists at UCSF have developed an AI-powered brain implant enabling a stroke survivor to communicate in both Spanish and English. The implant decodes brain activity into sentences, allowing the user to switch languages seamlessly, offering hope for more natural communication among bilingual individuals with paralysis.

Zuck Regains Cred

Mark Zuckerberg has regained favor in Silicon Valley by promoting open-source artificial intelligence. Developers appreciate that Meta’s AI system, released last summer, allows for free modification and reuse, contrasting with the closed models of Google, OpenAI, and Microsoft. This move aligns Zuckerberg with the hacker ethos and has earned him support from the tech community.

AI Writes More Convincing Restaurant Reviews Than Humans

A Yale study led by Balázs Kovács revealed that GPT-4 generates restaurant reviews more convincingly than humans. Participants misidentified AI reviews as human-written more often than not and expressed greater confidence in these misidentifications. This raises concerns about AI's potential to produce hyperrealistic fake content, affecting trust in various domains, including elections.

AI Models Have Favorite Numbers

AI models display surprising human-like biases in selecting random numbers. A study by Gramener showed that major AI chatbots, including OpenAI's GPT-3.5 Turbo, Anthropic's Claude 3 Haiku, and Google's Gemini, consistently prefer certain numbers and avoid others, mimicking human tendencies. This behavior highlights AI's reliance on past data rather than true randomness, underscoring the anthropomorphic traits AI can exhibit.

MIT Professor Predicts AI Won't Boost GDP as Expected

AI's impact on GDP and productivity may be overestimated, argues MIT professor Daron Acemoglu in a National Bureau of Economic Research paper. Skeptical of optimistic forecasts, Acemoglu estimates AI will only boost GDP growth by 0.93% to 1.16% over the next decade. Additionally, AI might not reduce inequality and could widen the income gap between capital owners and workers.