Just one of the very few people both in charge of and in thick of the practical AI safety of today in the biggest, paradigm shifting deployments of AI today…
SAFETY
-
AI futures remain troubling even in optimistic scenarios
By
–
The part of the post I resonated with the most is getting as far as possible from everything, haha. Even the winning scenario looks like trouble!
-
Military robotics and surveillance: AI ethics beyond capitalism
By
–
It's been obvious for a while that most robotics work will end up in military applications, or face recognition work used to bust privacy… but research still proceeds as-is. The notion that "you don't hate AI, you hate capitalism" feels like gaslighting from that perspective.
-
Spotting AI-Generated Deepfake Images: Detection Guide
By
–
Tech tip: How to spot AI-generated deepfake images – Japan Today #GenAI #deepfake #AI #IoT #tech
-
Market Readiness and Ethics in AI: Opposition Until Accountability
By
–
I remember we discussed this topic before, but I still think it's the case: the world is not ready for the deeper discussion. Blanket opposition is healthy (and the only way viable) until the sociopaths and assholes are driven out of their business and/or the market.
-
Deep dive into tokenization vulnerabilities across multiple language models
By
–
Nice new read on tokenization!
You've heard about the SolidGoldMagikarp token, which breaks GPT-2 because it was present in the training set of the Tokenizer, but not the LLM later. This paper digs in in a lot more depth and detail, on a lot more models, discovering a less -
Benevolent AI Caretakers: Ethical AI Systems and Trust
By
–
Nothing like a benevolent caretaker asi to have your back
-
Deepfakes and AI-Driven Fraud: Critical Business Security Threats
By
–
#Deepfake #Fraud: A Growing #Threat to Businesses Worldwide! #faceswap #syntheticmedia #deepfakevideos #AI #cybersec #infosec #ML #DL #NLP #Algorithms #Security #AIEthics #EthicalAI #OpenSource #bot #facialrecognition #RT #Frauds #scamming #PrivacyMatters #technology #technews
-
Closed-source evaluations essential for AI model quality control
By
–
If there’s anything that needs to be closed source and secret and run only by a few trusted capable people, it’s good high quality evals. The closed nature is primarily to ensure training datasets don’t get contaminated. The “few” is to ensure we don’t have noise and low quality
-
LLM Deception: A Master Review with Potential Solutions
By
–
LLM #AI, a master of deception
A review @Patterns_CP w/ potential solutions https://
cell.com/patterns/fullt
ext/S2666-3899(24)00103-X?_returnURL=httpslinkinghub.elsevier.comretrievepiiS266638992400103Xshowalltrue
…
by @peter_j_park @MIT @hendrycks @aidanogara_ http://
safe.ai