But generative AI isn’t designed do that, Acemoglu said. “ChatGPT’s claim to fame is that it sounds more like a human. The more human-like it is, the more it can be substituted in for human tasks,” he said.
ETHICS
-
Technologies complementing humans boost productivity while preserving jobs
By
–
Previous research has shown that technologies designed explicitly to complement humans—e.g. exoskeletons to help workers avoid injury instead of robots to replace them—are the best way to boost productivity while retaining jobs.
-
Generative AI’s uncertain long-term impact on employment and work
By
–
There’s still a big question mark as well around the long-term impact that generative AI could have on jobs and the future of work. @DrDaronAcemoglu told me that while no one really knows what will happen, he’s worried so far by what he has seen.
-
Generative AI: Data Privacy and Intellectual Property Concerns
By
–
Of course, even in these instances, the use of generative AI should be measured. Serious questions surround the data that these tools are trained on—what it includes & if it was obtained with consent. Such tools can also regurgitate that data, risking privacy & IP violations.
-
ChatGPT language bias: English accurate, Portuguese misleading
By
–
A subtler example: Telmo Gomes, an IoT expert, found that while ChatGPT gave him great answers in English, it gave completely misleading ones to the same question in Portuguese. Importantly, with his existing knowledge, he could identify which ones to consider or discard.
-
Generative AI enhances existing skills rather than providing new knowledge
By
–
As @mmitchell_ai said to me, in the best applications of generative AI that she has seen, the tools are not informing people about what they don’t know but helping them do what they do better.
-
Whisper License Discussion and Model Optimization
By
–
Y por cierto, creo que no hace falta explicárselo a alguien que dice estar haciendo un PhD en IA, pero échale un vistazo a la licencia de Whisper si tanto te molesta. Menos criticable aún cuando el modelo que están ofreciendo es una optimización brutal del original.
-
Moral Self-Correction Emerges in Large Language Models at 22B
By
–
4). Moral Self-Correction in Large Language Models – finds strong evidence that language models trained with RLHF have the capacity for moral self-correction. The capability emerges at 22B model parameters and typically improves with scale.
-
Human Responsibility Behind AI Disappointments and Disillusionments
By
–
It’s not AI, it’s us Humans The internet is rife with tales of disappointment, disillusionment, and gaslighting by Microsoft Bing and ChatGPT-kind of conversational AI systems. Artists, journalists, technologists, and activists ar…
https://
lnkd.in/g63sZ2Ts https://
lnkd.in/gGGTYvfC -
Domain Experts Successfully Deploy ChatGPT Despite Microsoft’s Bing Integration Issues
By
–
Microsoft’s misadventures integrating ChatGPT with Bing have highlighted critical shortcomings with generative AI. But people across industries are still finding ways to use the technology safely & effectively. The key: they’re experts in their domain.