Prolific
@Prolific
The ultimate human data platform to power world-changing AI and research. For help 👉 http://researcher-help.prolific.com / http://participant-help.prolific.com
Introducing the Human Intelligence Layer. Next-generation AI needs human intelligence baked into the very infrastructure of how it’s built. It’s critical to creating AI that’s capable, aligned, and truly useful in the real world. Read our vision ⬇️
Thank you for having me😊. Look forward to speaking at this webinar tomorrow -- I’ll share thoughts on the critical role of #Humans in shaping #AI and #LLM leaderboards, research, development, and alignment. Looking forward to brainstorming with others building more…
One day to go 🚨 Join @Cohere’s Oliver Nan, @UW's @huashen218, and Prolific’s Nora Petrova for “Why AI leaderboards miss the mark,” tomorrow @ 12:00 PM EDT. You'll improve your approach to LLM benchmarking and learn to measure model performance more reliably. Register now ⬇️
This is *the* paper to read this week. It covers an astonishing amount of ground on the persuasive capabilities of frontier AI - from scaling laws, to post-training, to the driving mechanisms of a persuasive advantage. Very proud of @KobiHackenburg + the team at @AISecurityInst!
Today (w/ @UniofOxford @Stanford @MIT @LSEnews) we’re sharing the results of the largest AI persuasion experiments to date: 76k participants, 19 LLMs, 707 political issues. We examine “levers” of AI persuasion: model scale, post-training, prompting, personalization, & more 🧵
ChatGPT Agent is here and GPT-5 on the way. @xAI's Grok 4 is high on the ARC-AGI benchmark. Superintelligence is in sight for @Meta, @OpenAI, and more. What do end users think? Our US representative data shows debate on AGI, while 77% say AI models’ trust and safety matter most.
Alignment is not only an evaluation or technical challenge, but also about understanding how AI interacts with human behavior in the real world. One of the great things about working with human data is that you can contribute to building, monitoring AND understanding real-world…
Today (w/ @UniofOxford @Stanford @MIT @LSEnews) we’re sharing the results of the largest AI persuasion experiments to date: 76k participants, 19 LLMs, 707 political issues. We examine “levers” of AI persuasion: model scale, post-training, prompting, personalization, & more 🧵
Congratulations @KobiHackenburg and team. Pleased to have supported the largest-ever study of AI persuasion with 76,977 taskers, with 19 LLMs deployed and 707 political issues evaluated on persuasiveness. Give this a read ⬇️
Today (w/ @UniofOxford @Stanford @MIT @LSEnews) we’re sharing the results of the largest AI persuasion experiments to date: 76k participants, 19 LLMs, 707 political issues. We examine “levers” of AI persuasion: model scale, post-training, prompting, personalization, & more 🧵
One day to go 🚨 Join @Cohere’s Oliver Nan, @UW's @huashen218, and Prolific’s Nora Petrova for “Why AI leaderboards miss the mark,” tomorrow @ 12:00 PM EDT. You'll improve your approach to LLM benchmarking and learn to measure model performance more reliably. Register now ⬇️
At the @AI_AInstitute Generative AI Summit in NYC, our VP of Product was on the panel with @forthepeople and @Wayfair discussing how leading companies are building practical, trustworthy AI with humans in the loop 🎤 Watch the full panel ⬇️ #NYTechWeek aiacceleratorinstitute.com/humans-in-the-…
The rise of LLM leaderboards has led to a drop in reliability. @Cohere_Labs already identified systematic issues on Chatbot Arena. Join @Cohere’s Oliver Nan (a first author of “The Leaderboard Illusion”), @UW’s @huashen218, and Prolific on July 22nd. Click to register ⬇️
LLMs like ChatGPT are making it harder for AI model builders to distinguish authentic human responses. Prolific CEO @Phelimb and VP of Product Sara Saab talk about adapting to constant challenges as a human data platform that prioritizes data quality. Watch the chat and demo 👇
On #AIAppreciationDay we celebrate the achievements in AI technology, many powered by human intelligence at the core. We're proud to support next-gen AI builders with representative human data, to empower AI that works for everyone. Discover the Human Intelligence Layer ⬇️