Audio should be easier. You could for example only transcribe voices that match a voice that has given permission in that voice. You’d still need AI to analyze whether it was AI generated but I suspect doing that on voice is easier than text.
SAFETY
-
LLM Bugs: Finding and Discussing Detection Methods
By
–
LLM bugs and how to find them Part of the joy of curating @aidotengineer is I get to bring some of the best people in the community that you only see online, to meet and teach in person. Several of our speakers are giving their first talks -ever-! Daniel is one of the most
-
US Robotaxi Companies Under NHTSA Investigation Despite Limited Mileage Data
By
–
All 3 remaining standing or limping US robotaxi companies under investigation by National Highway Traffic Safety Administration NHTSA. Together they have driven 70 million miles, less than 1% of daily 9 billion miles in the US, so it is hard to measure their safety performance.
-
Responsible AI Design: Prioritizing People and Planet Over Profit
By
–
Prioritizing People and Planet as the Metrics for Responsible AI: https://
standards.ieee.org/wp-content/upl
oads/2023/07/ead-prioritizing-people-planet.pdf
… From @IEEEorg – Where you don’t begin design of tech asking what metrics can we use to put people and planet first, you’ll prioritize maximization of fiscal growth by default vs P&P. -
Robots Need Onboard Perception Despite Cloud Information Sharing
By
–
Nope! I'm saying that being able to get information from other places (think congestion info which current auto nav systems get from the cloud) or to contribute to the whole fleet when an indiv robot observes something. They still need onboard perception, control, and tasks.
-
Balancing AI Development with Comparable Safety Investment Resources
By
–
I am not "blindly opposing AI". I think it can be tremendously beneficial in areas like healthcare.
I am simply advocating that as we put huge resources into advancing AI, we put comparable resources into ensuring it is safe -
RLHF Hard-Coding: How AI Models Learn Topic Avoidance
By
–
Fascinating! I did find a prompt that didn't reference it at all, does it mean that the topic is too far away to Golden Gate Bridge, or that the question/answer has been hard-coded or strongly RLHF'd?
-
AI Bullshit: Shit-talking as critical praxis in AI discourse
By
–
AH: (singing) Slogging through bullshit, we’ll never relax as shit talking AI is ridicule as praxis. /end alt text
-
Combat AI Bias: Building Fair and Accountable Systems
By
–
This article delves into the origins and consequences of #AIbias in various sectors and offers actionable steps for organizations and individuals to build AI systems that are #fair, #accountable, and #transparent. https://
forbes.com/sites/bernardm
arr/2024/05/24/building-responsible-ai-how-to-combat-bias-and-promote-equity/
…