In an effort to enhance AI governance and model trust and transparency, SAS has announced the launch of new AI products and services, including the unveiling of model cards and an AI Governance Advisory service.

In launching the new products, SAS aims to assist organisations in navigating the often turbulent AI landscape, mitigating risk and promoting the confident pursuit of AI goals.

Model cards, described as “nutrition labels for AI models,” are regarded as instrumental tools across the AI life cycle. SAS VP of Data Ethics Practice, Reggie Townsend, underlined how organisations are excited about the concept of AI but remain wary of its best utilisation. Townsend stated, “They're asking good questions about responsible and ethical AI. Our goal is to answer these questions with the right tools and guidance, based on decades of experience, while ensuring AI integration in ways that increase profitability and minimise unintended harm.”

The new model cards feature is set to debut in SAS Viya in mid-2024. SAS plans to autogenerate model cards with content straight from SAS products, lifting the generation burden from individual users. Not only this, but SAS Viya’s existing framework will now also support open source variations, commencing with Python models. The model cards will provide insights into factors like accuracy, fairness, and model drift, revealing crucial governance details. They will also define intended and out-of-scope usages, revealing limitations that are expected to gain importance as transparency and model auditing transform into regulated business operations.

Research Director at IDC, Eric Gao, recognises that SAS is adopting a prudent approach, focusing on the practicalities and challenges of AI deployment across industry sectors. He commented, “Model cards will be beneficial for monitoring AI projects and fostering transparency."

To address growing customer concerns about using data in productive and safe ways amid the spread of AI, SAS is also launching the AI Governance Advisory. This value-added service assists customers as they navigate their respective data and AI landscapes. It fosters reliable and distributed decision making, encourages better data usage accountability, enables competitive advantages from future compliancy, and boosts brand value by proactively addressing potential societal and environmental impacts.

Financial institution in Central and Eastern Europe, PZU Insurance of Poland, has already appreciated the benefits of the AI Governance Advisory. Marek Wilczewski, Managing Director of Information, Data, and Analytics Management at PZU, reflected on how their dialogues with SAS have opened up an understanding around potential blind spots that could have caused complications for PZU's customers and business. He added, "We now grasp the significance of having more perspectives as we commence with AI projects."

Highlighting their commitment to ethical AI, SAS also announced that industry veteran Steven Tiell has been appointed as the Global Head of AI Governance. Tiell previously led the global data ethics and responsible innovation practice at Accenture and was the former Vice President of AI Strategy at DataStax.

Last year, the US National Institute of Standards and Technology (NIST) launched an AI Risk Management Framework, which is now being utilised by SAS to shoot for trustworthy and responsible AI in the absence of formal regulations. The SAS Trustworthy AI Life Cycle workflow is built onto NIST’s recommendations to facilitate adoption by organisations with the aim to generate a production model with transparent due diligence documentation showing that the model is fair and processes cause no harm.