(Europe) Heather Dawe, UST UK Head of Data, speaks with Kellie de Leon, Treasure Data Senior Director of Content Marketing, in a video interview about the pros and cons of black box AI solutions, the AI assurance ecosystem, and building tech stacks to avoid technical debt and ensure controlled evolution.

Speaking on the issue of black-box AI solutions, Dawe says there aren’t enough data scientists in the world, leading to the need for black-box technologies that are easy to use for nontechnical people. She mentions, however, there is a chance that the person developing the black-box model may need help understanding the risks associated with it.

Dawe adds that in the ideal scenario, data scientists work closely with the business team and help them develop that model and understand the issues.

Regarding the importance of AI assurance ecosystems, she says it is one of the most needed things across the industry. Dawe adds that these frameworks should help people confirm that their AI solutions are unbiased and safe to use across industries like health care and manufacturing. They will aid innovation within a framework and appropriate limits.

Next, Dawe highlights how businesses should approach their tech stack to support future evolution, noting the following focus areas:

• Recognize that implementation can be successful at scale in many ways using the technology available on the market. There can be more minor add-ons that will enhance the data processes.
• Avoid lock-ins because huge amounts of technical debt can burden businesses.
• Recognize that these things evolve and are evolving at pace.
• When a change takes place, consider the impact of that change and ensure controlled evolution.

CDO Magazine appreciates Heather Dawe for sharing her insights and data success stories with our global community.

See more from Heather Dawe

It’s Still a Challenge to Deploy Machine Learning Models at Scale

AI Is Nowhere Near Sophisticated Human Brains