“AI needs to do whatever i ask” and “i asked the AI to be sexist and it was, look how awful!” are incompatible positions. somewhat surprised by the number of people who hold both.
SAFETY
-
Constitutional AI: Moving Beyond Researcher-Defined Constitutions
By
–
In our paper we used an ad hoc constitution drafted purely for research purposes. Ultimately, we think constitutions shouldn’t be just defined by researchers in isolation, but by groups of experts from different disciplines working together.
-
Constitutional AI: Making Implicit Principles Explicit in AI Systems
By
–
While the name “Constitutional AI” may sound ambitious, we chose it to emphasize that powerful, general-purpose AI systems will always be operating according to *some* principles, even if they are left implicit, or encoded in privately held data.
-
AI Progress and the Growing Black Box Problem
By
–
The good news is we can do more and more with AI. The bad news is we understand less and less how it does it.
-
AI Learning Trust and Hype: Key Challenges Ahead
By
–
And this is a good summary of the key points: #ai mimics people's mistakes, needs to learn to learn vs being trained, needs to focus on trust! And I will repeat again, I only fear the hype! Thx @constellationr for giving us the platform. @AndyThurai was a great guide for us!
-
What if smaller, sparser models all fail?
By
–
there is ongoing research toward smaller, sparser, human-efficient models, but what if they *all* fail. scary thought.
-
Public vs Proprietary Datasets in AI Model Training
By
–
Thanks Alex. I like the papers that do this, but I also have some concern when this is done on a dataset that is proprietary and only Google has access (JFT). I would like to see a version that is pretrained on LAION. This still has privacy issues, but is at least all public.
-
IBM Research Explores Deep Learning Methods to Reduce AI Bias
By
–
#AI bias is more than just unfair – it can amplify social inequalities and create distrust in technology. Using #deeplearning, @IBMResearch are exploring ways to reduce this bias in large pre-trained AI models: https://
ibm.co/3BxXObP -
Private ML Benchmarks and Privacy-Focused Evaluation Standards
By
–
The second is to re-focus towards benchmarks that are more appropriate for private ML. We now understand that public data can help for private CIFAR-10 and ImageNet classification, which is great. But maybe we should move towards settings where privacy is more important. 13/n
-
Privacy-Respecting Public Pre-Training Datasets for AI Models
By
–
So where do we go from here? We conclude with a number of suggestions for the field. The first ones focuses on making sure we have public pre-training sets which are truly privacy-respecting. Can we make such a dataset/model with comparable utility to what people use now? 12/n