© Copyright SUBEX 2023. All Rights Reserved.
Why is there a need for measurement of Trust in AI?
- Assesses how resilient the model is to changing datasets over time
- Evaluates if the model’s explainability is on par with real-world understanding
- Assess the model’s fairness and if a model is biased toward any certain attribute
- Evaluates the model’s robustness to various adversarial attacks
YOUR SIGN-UP IS GETTING READY
Microsoft AI Chatbot “Tay" was shut down within 24 hours of its launch due to a series of lewd and racist tweets.
Amazon’s secret AI recruiting tool was biased against women and would prefer male candidates for the job.
Facial recognition has been fooled using a 3D-printed mask to authenticate the Facial ID systems
Uber’s self-driving vehicles failed to classify the object as a pedestrian while test driving in US, which led to the fatality.
AI struggled with image recognition when Google’s racist algorithm tagged people as gorillas.
Tesla recalls many vehicles because AI lets them roll through stop signs which can cause fatal accidents.
- According to a KPMG survey, more than 50% of executives stated a lack of AI ethics guidelines as a barrier to adoption and progress.
- According to Gartner, “Responsible AI helps achieve fairness, even though biases are baked into the data; gain trust, although transparency and explainability methods are evolving; and ensure regulatory compliance while grappling with AI’s probabilistic nature.”
- Forrester predicts that the market responsible for responsible AI solutions will double in 2022.
- Gartner predicts that by 2023, all personnel hired for AI development and training work will have to demonstrate expertise in responsible AI.
Our Solution Framework
We built a model auditing tool “AI Trust Calculator” to measure and quantify “Trust” on key pillars of our solution framework. For this innovation and contribution of this to TMF, we were awarded Best Catalyst Project under Leadership of Sustainability, where TMF recognized this framework aligning with United Nations Sustainable Development Goals.
Note: The AI Trust Calculator is only to perform ML model diagnostics and generate vulnerability report. The assessment is based on our proprietary tools and powered by HyperSense AI. There is detailed methodology of the assessment performed. The recommendation report can be requested post assessment. Please reach out to firstname.lastname@example.org for any questions or recommendations on your model assessment.