Data Science Consultancy, Tessella, Calls for Professionalization of AI to Address High Failure Rates and Lack of Trust

Data Science, Tessella, AI

Tessella, Altran’s World Class Center for Analytics and AI, has released a whitepaper recommending a professional framework for delivering successful data science and AI projects, which will increase trust and reduce sky-high failure rates.

Tessella ​– which has 40 years’ experience in data science consulting and delivers AI projects for the likes of BP, GSK, and Unilever – argues that professional industry standards need to be established in AI. They now share their own AI development governance framework to kickstart these changes.

Read more: Shanghai HiCore Tech Announced Partnership With MIT on Anti-Fraud AI Lab

“There’s an obvious analogy to the early days of software development,” says Matt Jones, Lead Data Science & Analytics Strategist at Tessella. “Many early products were prone to bugs or high levels of technical debt once deployed. Eventually, we learned that long-term success needed proper processes. Software is now a highly professional global industry, with standards for programming languages, development methodologies, project management frameworks, and certified industry accreditation.”

Read more: Blockchain Will Disrupt Enterprise Like the Cloud

“Many present-day AI projects have high failure rates, including those developed by leaders in the field; from Amazon’s ‘sexist’ recruitment tool to IBM Watson’s shortcomings in healthcare. AI needs to go through a similar professionalization process to software to tackle this problem and build user trust that it will deliver as required.”

Also read: Human Intelligence Can Fix AI Shortcomings

Tessella’s data science framework, RAPIDE, is many years in the making. It is mature, robust and provides a firm reference for others. “We hope that putting this out there will help others, but also build momentum towards creating industry standards to reduce failures of data science and AI projects,” says Jones.

Tessella’s RAPIDE ensures that AI projects: are business Ready; only use data that passes Advanced screening; look beyond chance correlations to Pinpoint the real factors driving outcomes; identify and evaluate multiple AI models, methods and toolset options; develop models where trust is the equal of raw predictive power and evolve their capability upon contact with the real world.

Also read: XtremeLabs Partners with GreenBridge Computing to Deliver Low Cost, Green Solutions for Training and Learning

This multi-stage approach allows rapid experimentation to quickly reduce many ideas down to the most viable and successful ones, and spot dead-ends before costs spiral. This is essentially the famous ‘fail fast’ approach which has been integral to the success of today’s tech giants.

Also read: Plataine and Marubeni Aerospace Sign MoU to Offer IIoT Solutions

“Now is the critical time to create a rigorous professional framework for data science and AI, articulating and crystallizing best practice and ensuring we can trust AI to deliver as expected,” concludes Jones. “Otherwise, history will repeat itself, and we’ll make the same mistakes as we did at the start of previous technology revolutions.”

Must read: Deloitte’s Heat Launches Artificial Intelligence Practice ‘Heat AI’

Previous articleKoreConX Announces First Fully Compliant Digital Securities Offering for Middle East & Africa
Next articleTwo Unexpected Artificial Intelligence Adopters Industries