Learn how differentially private stochastic gradient descent (DP-SGD) can be applied to train ad prediction models privately with more improved model utility than previously expected, all while reducing computation and memory overhead. Read more → https://
goo.gle/3VUTlbn
ETHICS
-
Differentially Private SGD Improves Ad Model Training Efficiency
By
–
-
Programming’s Balance: Creativity Versus Addiction
By
–
Perhaps. The paradigm of programming involves some parts that are the highest expression of the human mind, and some parts that are simply a muscle response to instant gratification. And in the balance there may be creativity, on the one hand, and merely addiction on the other.
-
Will AI Seduce Humans Away From Their Own Instincts?
By
–
It's a beautiful thought, beautifully expressed. And one corresponding question is whether the human, in the process, will be seduced away from their own instincts into serving the relentless feedback loop of the computer, despite humans' better nature. It's an open question.
-
AI Feedback Loops: Mutual Manipulation Between Human and Machine
By
–
Remember that AI becomes a feedback loop. The human can become just as much a thing to be manipulated by the machine, even when the human is nominally in the role of deciding the creative direction. The dialectic effects both parties, human and machine.
-
Citations and Fact-Checking: Foundations of AI Accountability
By
–
Citations are a way to ensure that non-accurate results can be fact-checked. Citations inspired (albeit in a different way) the Google PageRank Algorithm.
-
Search Accuracy Issues in LLM Entity Resolution Systems
By
–
Results are not always accurate. For example, a search about “what are LLMs” struggles with entity resolution with the law degree and large language models. Feedback for negative results could help fix this. pic.twitter.com/w1JtYehRup
— Perplexity (@perplexity_ai) 7 décembre 2022Results are not always accurate. For example, a search about “what are LLMs” struggles with entity resolution with the law degree and large language models. Feedback for negative results could help fix this.
-
AI crossing from helpful to problematic articulated clearly
By
–
Yes!! You put into words something I’ve been struggling to articulate about when AI crosses over from helpful to problematic.
-
Cautious approach needed: training data privacy considerations
By
–
The conditions necessitate a cautious approach. Plus, the training data isn’t the personal or private information of individuals.
-
Large Language Models in Drug Industry: Measurable Benefits and Regulation
By
–
This is what I like about this particular application of large language models. The drug industry is highly regulated, and there are clear ways to measure whether the output of a protein-language model is doing more harm than good.
-
Understanding AI Limitations Should Not Be Penalized
By
–
Knowing your limitations should not held against you