Interesting that it was concerns of an algorithm called Q* that fed in part into Sam Altman's ouster, per Reuters. Some of my sources say they would not consider Q* a big breakthrough. Keen to hear from other researchers in the company about their opinion. https://
reuters.com/technology/sam
-altmans-ouster-openai-was-precipitated-by-letter-board-about-ai-breakthrough-2023-11-22/
…
SAFETY
-
Q* Algorithm’s Role in Sam Altman’s Ouster at OpenAI
By
–
-
Project Relate: Personalized Speech Recognition for Non-Standard Speakers
By
–
Project Relate is a beta Android application that offers personalized speech recognition for non-standard speakers. Meet Alan, travel enthusiast and retired railway worker, who uses Project Relate to break communication barriers and make his voice heard. https://t.co/B4Vb5SUQ4B pic.twitter.com/nkZ8qxnWt6
— Google AI (@GoogleAI) 22 novembre 2023Project Relate is a beta Android application that offers personalized speech recognition for non-standard speakers. Meet Alan, travel enthusiast and retired railway worker, who uses Project Relate to break communication barriers and make his voice heard. https://
goo.gle/47mbPHb -
Why Robots Cannot Pass CAPTCHA Tests
By
–
Why cant robots pass captcha tests? #innovation #tech #captcha #AI #ML #Robotics #robots #ChatGPT #chatbots #chatgpt4 #chatgpt3 #GenerativeAI HT @alvinfoo @ipfconline1 @JimHarris @PawlowskiMario @Hana_ElSayyed @dinisguarda @bimedotcom @Shi4Tech @digitalhealthxx @stratorob
-
No Bug: Designed Harm and Institutional Denial
By
–
“So what was the bug?” Ah, you see, that’s the thing. There was no bug. Everything happening was designed to happen. No party admits fault. No official party even institutionally believes that harm happened.
-
OpenAI Leadership Returns to Drive AI Innovation and Safety
By
–
I’m excited to see @sama
, @gdb and the gang back together at @OpenAI
. They have undeniably changed the world over the past year, and I expect them to continue setting an impressive pace in AI for building, shipping, and innovating on safety. Nature has healed -
AI Security: Explore Research Complexities and Interactions
By
–
If the security of AI interests you, or if you're simply curious about the complexities of these interactions, watch the full video on this research for a deeper perspective. Visit https://
youtu.be/tmKgL0eELKM -
AI Safety: Understanding Defenses and Digital Threats
By
–
Understanding AI's defenses and weaknesses is essential for its safe integration into daily life. Initiatives advance knowledge and shape measures to protect against digital threats.
-
Prompt Hacking: Understanding AI Vulnerabilities and Defense Mechanisms
By
–
Prompt hacking involves using ingenuity to make AI breach its ethics, leading to harm. This challenge shows the need to grasp such interactions to boost AI's defenses against abuse.
-
Contest Creates 600K Adversarial Prompts for AI Security Research
By
–
We made a contest that yielded a key data set of 600,000+ adversarial prompts, sparking a taxonomy of hacking techniques and advancing AI security insights.
-
Prompt Hacking: Security Issue in LLMs and Solutions
By
–
'prompt hacking' lets one access events uninvited, posing a big issue to models like ChatGPT or any LLM, which we tackled with @learnprompting and colleagues from @mila ( @jerpint ), @ChengleiSi from @stanfordnlp
, and @towards_AI
.