IBM’s Ginni Rometty Calls for Transparency and Ethics in AI

Artificial intelligence has entered the mainstream. With the powerful technology beginning to show up in everything from toasters to cars, people and organizations alike are growing concerned that there may not be enough oversight to prevent AI from becoming a threat to jobs and privacy.

IBM’s CEO, Ginni Rometty, is well-aware of these concerns. At a talk reported on by ZDNet, Rometty noted that, “Cognitive systems will not realistically attain consciousness or independent agency.”

Why IBM Had a Stake in the Debate

IBM wasn’t speaking about AI from the sidelines. Its Watson system had become famous in 2011, when it beat Jeopardy! champions Brad Rutter and Ken Jennings and won the US$1 million first prize. By the time Rometty set out these principles, Watson was at the center of IBM’s pitch to businesses, and the company was selling it into areas like health care, where questions about trust and data are especially sensitive.

That gave IBM a clear reason to get ahead of public worry. If customers and governments came to see AI as a threat, a company betting so much on it would feel the effects first.

IBM’s Three Principles

Still, IBM has created three core principles it will adhere to when it comes to AI, including:

• Purpose: IBM’s development of AI is focused on “augmented human intelligence,” not artificial intelligence
• Transparency: The company plans to publicize its AI work and the data it uses
• Skills: IBM acknowledges that AI will create change and that deployments of the technology must consider the new skills that people will need

Each principle answers a specific fear. “Augmented” intelligence is a promise that AI will help people make decisions rather than replace them. Transparency speaks to the worry that AI systems are black boxes, trained on data nobody can inspect. And the skills principle is an admission that jobs will change, even if IBM frames it as a reason to retrain workers rather than a threat.

For more, read the full article on ZDNet.

What Happened Next at IBM

Rometty led IBM for three more years. In January 2020, IBM announced Rometty would step down as CEO, and Arvind Krishna was named as the next CEO. Krishna took over as CEO in April 2020, and Rometty retired from IBM at the end of that year, when Krishna was also elected chairman. Rometty had been the first woman to lead the company and spent nearly 40 years there.

Watson’s story was rockier. The health care business IBM had built around it never became the success the company hoped for. In January 2022, IBM announced that Francisco Partners would acquire the healthcare data and analytics assets that were part of Watson Health, including products such as MarketScan and Micromedex.

IBM then rebuilt its AI offering for a new era. In May 2023, it announced watsonx, a platform for businesses to train, tune and deploy AI models, including foundation models. Notably, the launch included a toolkit for governance of AI, which shows how the transparency theme from 2017 carried into IBM’s newer products.

How the Debate Has Moved On

Rometty’s comment that AI wouldn’t “realistically attain consciousness or independent agency” looks different today. The arrival of generative AI tools that can write, code and hold conversations has made AI far more visible to ordinary people, and talk of AI “agents” that act on a user’s behalf is now common across the industry.

The core questions she raised haven’t gone away, though. People still want to know what data AI systems are trained on, whether AI will replace their jobs, and who’s responsible when an AI system gets something wrong. Those are the same issues IBM’s purpose, transparency and skills principles were meant to address.

What It Means for Businesses Using AI

IBM’s 2017 principles are a useful checklist for any company adopting AI. Be clear about what the system is for and where a person stays in charge. Be open about the data it relies on. And plan for how roles will change, including the training people will need. Those steps won’t settle every concern, but they’re a good starting point for earning trust from customers and employees.