And finally, you might have seen our collaboration with Anthropic’s Societal Impacts team on Feature Steering, also published today:
ETHICS
-
Academic Integrity and AI Ethics: University Accountability
By
–
I worked tirelessly to expose him and received how many attacks in the process. I refused to appear at @UniofOxford as long as they have the likes of him, and now even they could no longer be associated with his so-called institute.
-
Ethical concerns about promoting controversial AI figures
By
–
And not only do you have an article praising a eugenicist I've written about as a notable figure to follow, but you casually add his name next to mine?
-
Controversy Over AI Researcher’s Statements Raises Ethics Concerns
By
–
@chicagotribune why would we follow the voice of a eugenicist who calls us the N word, "blacks are more stupid than whites" and worried about stupid people reproducing too much AKA Nick Bostrom? And how dare you have his name next to mine?
-
Critical Voices in Technology Ethics and AI Governance
By
–
This is something in @chicagotribune
. Spot the issue. "let’s remain critical about technology & educate ourselves by following the works of such notable voices in this field as Nick Bostrom, Mark Coeckelbergh, Timnit Gebru and Shoshana Zuboff." -
Research on AI Steering Methods for Safer Model Outputs
By
–
We hope our preliminary findings inspire further research on steering methods for safer, more reliable model outputs. Read the full research blog for detailed results, insights, and limitations:
-
Feature Steering Reduces AI Bias Across Nine Social Dimensions
By
–
Finally, we discovered a feature that significantly reduces bias scores across nine social dimensions within the sweet spot. This did come with a slight capability drop, which highlights potential trade-offs in feature steering.
-
AI Gender Bias Mitigation Unexpectedly Increases Age Discrimination
By
–
However, we also observed unexpected "off-target effects". For example, to our surprise, we found that dialing up the "Gender bias awareness" feature also increased age bias.
-
Finding Sweet Spot for Steering Model Social Bias Features
By
–
We found a "steering sweet spot" where feature steering influences model outputs in intuitive ways without significantly degrading the model’s capabilities. Surprisingly, all 29 social bias related features we tested shared the same sweet spot.
-
Feature Steering Controls Social Bias in AI Models
By
–
Next, we found that feature steering can indeed increase or decrease various forms of social biases in targeted ways. For example, dialing up the "Gender bias awareness" feature significantly increased the gender bias scores in our evaluations.