What’s It About?
LinkedIn has introduced a new feature called Crosscheck that allows users to compare different AI models directly against each other. The distinctive aspect: the identity of the models remains hidden at first, ensuring an unbiased evaluation. Currently the feature is available to Premium subscribers in the US, with expansion to additional user groups and countries planned.
Users enter a prompt and receive two different responses from various AI systems — such as those from OpenAI, Anthropic, or Google. Only after they have chosen the better answer is it revealed which model produced it. Unlike many paid AI services, Premium members can conduct an unlimited number of comparisons.
Background & Context
The concept of blind testing AI models is not new — similar approaches already exist on platforms like LMSYS Chatbot Arena. What is new is LinkedIn’s integration of this feature into a professional networking context. This allows working professionals to directly test which models perform particularly well for specific professional tasks. The feature is labeled as an Early Product, indicating that LinkedIn plans to integrate further models and functions. The blind test mechanism is designed to help eliminate brand preferences and enable objective evaluation.
While other platforms often offer access to various AI models only for separate fees or with usage limits, LinkedIn differentiates itself through unlimited use within its Premium subscription. This could establish the professional network as a testing environment for AI technologies while simultaneously delivering valuable data about the preferences of professional users.
What Does This Mean?
- LinkedIn creates a low-barrier opportunity for professionals to practically compare different AI models without having to subscribe to separate services.
- The blind test approach could lead to more objective evaluations and show whether more expensive or better-known models actually perform better in practice.
- For businesses and HR departments, the collected data on AI model performance in professional contexts could be relevant for product development and partnerships.
- For AI providers, a new competitive channel is emerging where their models are directly compared with the competition — which could increase the pressure to innovate.
- The initially limited availability to US Premium users indicates a phased rollout, with LinkedIn likely wanting to collect feedback and optimize the feature first.
Sources
- Test and Compare AI Models with LinkedIn (Computerwoche)
- You Can Now Test and Compare AI Models on LinkedIn (Computerworld)
- LinkedIn’s Crosscheck Feature Lets Premium Subscribers Test AI Models for Free (Engadget)
- LinkedIn’s New Tool Lets You Test the Outputs of Various AI Models (Social Media Today)
- LinkedIn Wants You to Judge AI Models Head-to-Head Style with Crosscheck (Neowin)
This article was created with AI assistance and is based on the cited sources and the language model’s training data.
Further Reading: From Text Generator to Digital Employee: How AI Is Changing the World in Four Stages