What’s new: Faster AI evaluations, enhanced messaging, popular screeners, and more!

Welcome to our latest product update! We’ve released several updates to our platform that make AI evaluations and research much easier on Prolific. Read on to learn about what’s new, including AI taskers, AI Task Builder, messaging enhancements, and some exciting AI events.

Introducing AI Taskers: Quality human data for better models

Access a dedicated pool of AI taskers who understand what you need, deliver reliable outputs, and help you build better models faster. Available now on your study dashboard or via our API.

• Instant access, immediate results: Connect with pre-verified AI talent without the screening hassle. Start collecting high-quality data from day one and accelerate your development cycles.
• Superior quality you can trust: Our AI taskers exceed rigorous performance benchmarks in model evaluation, comparative reasoning, and information verification. This ensures your models learn from the best.
• Flexible scaling that grows with you: Seamlessly adjust your participant pool up or down as your project needs change. Maintain exceptional data quality without interruption.

Level up your AI evaluations with AI Task Builder

AI Task Builder early access is now available to all Prolific users! It's already powering evaluations at scale, from fact-checking AI-generated content to evaluating translations.

We’ve just launched three powerful new features to enhance your annotation workflows:

• Checkbox with conditional free-text: Capture the "why" behind the data by prompting annotators for explanations when they select specific options. Ideal for identifying patterns in model hallucinations or categorizing output quality issues.
• Dynamic task grouping: Organize annotation tasks by automatically grouping related datapoints using your own identifiers in CSV datasets. Perfect for handling variable evaluation criteria or unevenly distributed examples.
• Support for larger datasets: Upload files up to 5GB (equivalent to 500,000 conversation transcripts or 625 million words). Run evaluations across diverse test cases without splitting datasets or managing multiple runs.

Ready to improve your AI models and products with human feedback? See AI Task Builder in action with our interactive product tour. Or dive right in by accessing AI Task Builder in your Prolific dashboard.

Team messaging to help you get quality data, faster

Clear communication with participants is key to getting the highest-quality data. Our new messaging features give you unprecedented control and efficiency in participant interactions:

• Participant group messaging: Reach thousands of participants with a single message. Now available to all users, this feature transforms how you deliver announcements, clarify instructions, and send reminders.

• Workspace messaging: Accelerate your participant feedback loops with a single shared inbox for your workspace. Now, every workspace member can see and respond to messages with participants and participant groups. This removes communication bottlenecks and improves data quality.

  Note: Workspace messaging is currently available to select workspaces. If you want access, register your interest here and we’ll be in touch. This feature can’t be reversed once implemented. So, we recommend reviewing your team’s privacy requirements and establishing team communication guidelines before activation.

To learn more about these features, read our help center articles on participant group messaging and workspace messaging.

Instantly find the most popular screeners

Prolific’s loved for its hundreds of free audience filters, but some are more popular than others. So we’ve made it really easy to find the screeners you need the most, based on data from 2.5 million studies.

When finding your participants, select the “Popular” tab to see the filters used the most, e.g. Age, Language, and Sex. Go to “All screeners” to find the full suite of options, e.g. pet owners, environmental beliefs, or weekly exercise.

Improved participant experience

The participant experience is now smoother than ever, thanks to some critical upgrades that improve dashboard loading times and reduce reservation errors.

• Faster study access: Study dashboard loading times have been slashed from up to 8 seconds to 3 seconds or less. This improves study uptake and participant engagement.
• Seamless study reservations: Our refined study distribution rules have substantially reduced reservation errors, creating a frustration-free experience for participants.

These improvements work together to enhance the participant experience on our platform.

Run faster research with our new partner tool, Versive

Teams often skip research because they “don’t have time”, leading to riskier product outcomes. Versive changes this, with AI-assisted studies and analysis that gives you interview-quality insights at the speed of a quant survey.

Coupled with Prolific's rapid participant responses, Versive and Prolific make the perfect match for faster research.

To use Prolific and Versive together, connect Versive as you would any other survey tool—with just a link. Learn more about Versive.

In other Prolific news

Workshop: Learn how to find and qualify exceptional AI annotators

17th March at 2pm GMT (UK)

Join us in London at AI UK for our in-person workshop, The Human Element in AI development: Developing a Pipeline to find Super Annotators.

Learn about the practical challenges and solutions of qualifying skilled taskers for AI model training and evaluation. We’ll detail Prolific’s approach and you’ll get the exclusive chance to try sample assessments from our qualification process.

Webinar: Learn how to integrate human feedback to evaluate your AI products

20th March at 4pm GMT / 8am PST

Join our practical webinar on human evaluation for AI development. You'll learn when to use human feedback and how to overcome common implementation challenges. Plus, you'll see Prolific's AI Task Builder in action on an evaluation use case.

Perfect for product managers, ML engineers, and data scientists looking to enhance their AI evaluation processes and build better products.