AI will have an increasing impact on our daily lives. Amidst all the enthusiasm around artificial intelligence, SAS draws attention to a fundamental question: how do we ensure that AI is trustworthy?

Reggie Townsend asks the audience to stand up in the middle of his keynote. He then asks who in the room has already had a meaningful relationship and whose trust in a relationship has already been broken. Anyone who can answer one of these questions with “No” can sit down, but after two questions, almost the entire room is still standing.

Townsend clarifies what he wants to say: “Trust is the foundation of any relationship. It’s a feeling that’s deeply rooted in our society. People also want to feel secure in technology. That’s one of the big challenges for the tech sector, because how do you create trust in something that you can’t physically see or hear? You can’t force trust, you have to earn it.”

As VP of the Data Ethics Practice at SAS, Townsend clearly cares about this topic. Trust between people and technology will also be an important theme during the SAS Innovate conference in Amsterdam. SAS wants to be a pioneer in developing “reliable AI” that serves people. This is already mentioned in the foreword by Gavin Day (Executive VP) clearly: “I’ll say it now, but the topic of reliable AI is discussed in all lectures today.”

Rapid growth

Developments in AI are therefore progressing very quickly. The current craze for generative AI has accelerated development and adoption: suddenly every company wants to do something with AI. IDC predicts that the artificial intelligence market will grow at a compound annual rate of 18 percent through 2026; According to Gartner calculations, the market will then be worth 135 billion US dollars. PwC goes one step further and predicts that artificial intelligence will contribute $15.7 trillion to the global economy by 2030.

Alice McClure (Director of Technology) sees the current economic climate as breeding ground for the growth spurt of AI: “Everyone feels the pressure on staff today. Organizations need to do more with fewer resources. AI can then provide a solution to increase productivity.”

People want to feel secure in technology. That’s one of the big challenges for the tech sector, because how do you create trust in something that you can’t physically see or hear?

Reggie Townsend, Vice President of Data Ethics Practice SAS

Doing well

Rush and urgency are rarely good, and that certainly applies to the development of artificial intelligence. Models that are thoughtlessly developed, trained, or implemented can harbor a hidden danger: bias. Bias is often invisible to a system’s designers, but can seriously skew AI-driven decisions made using that system, Day warns. And when those decisions affect people, there can be undesirable consequences.

Of course, no sane scientist would take it upon himself to create a system that would intentionally harm people. Moreover, they would also be severely punished by society, as a study by Capgemini shows.

When customers feel that a company’s moral compass is pointing in the right direction, they buy products faster and are also more loyal to that company. This trend also works in the opposite direction when customers have doubts about your good intentions. And as if that wasn’t reason enough to get involved in charity, the upcoming AI law will impose stricter requirements on the security of AI technology for companies.

In his speech, Townsend emphasizes that ethical and economic goals are not contradictory. “As a company, you can perfectly aim for profit and at the same time do good for humanity, they don’t have to be mutually exclusive. People generally have a good idea about companies. This is an opportunity you should seize. AI should reflect the best of humans in everything. Our motto is therefore Do good and do it well.”

As a company, you can perfectly aim for profit and at the same time do good for humanity, they don’t have to be mutually exclusive. AI should reflect the best of humans in everything.

Reggie Townsend, Vice President of Data Ethics SAS

Four steps to reliable AI

Trustworthy AI means more than just good intentions, it also needs to be implemented in reality. According to Capgemini’s research, this still appears to be a major stumbling block for many companies. 77 percent of business leaders have doubts about the transparency of AI systems and the use of AI has already led to ethical dilemmas in nine out of ten companies. The pressure to jump on the bandwagon is an increasingly common reason.

To help companies find the right path, SAS has created a guide to putting the principles of trustworthy AI into practice. It consists of four parts, each of which has the same weight in ethical considerations. Sufficient attention must therefore be paid to each component in order to develop a successful and responsible AI strategy.

overview

According to SAS, it all starts with an overview of the big picture. Companies must first ask themselves the “why” question before they start building an AI system. Why do you as an organization want to work with AI? Does the technology align with your values ​​and goals? Ultimately, the AI ​​strategy should complement your broader business strategy.

Check over

After the overview comes the supervision. In this phase you should not only ask yourself what you can do with AI, but above all what you are allowed to do with it. To do this, you need to examine what regulations your organization is bound by and whether you have the right resources and procedures internally to comply with them.

But SAS insists on looking beyond legal obligations. Every organization should conduct an internal analysis of any new technology to identify and manage potential risks. A risk reduction plan must then also be drawn up in order to be able to intervene in the event of unforeseen risks.

platform

The risk analyzes have uncovered potential vulnerabilities that could prevent responsible use of AI. Now it’s time to deal with it. If necessary, take a critical look at your data management practices and refine policies or implement new ones as needed. Make sure they are practical, doable and easy to implement. The system of “model cards”, for example, enables technology companies to communicate transparently about their systems to the outside world.

From the data management you can then view your entire IT ecosystem. Organizations must dare to evaluate both internal software tools and those of external suppliers against their own principles. Consistency across the entire ecosystem is a key requirement for an effective approach to trustworthy AI.

You need to know where your technology is emerging and how it can be used to ensure that it doesn’t end up causing harm to people.

Reggie Townsend, Vice President of Data Ethics SAS

Culture

To achieve this cohesion, one crucial component must not be forgotten: culture. Innovation is not a purely technological, but also a human process. A technology is only as secure as the way users interact with it, and that’s certainly the case with AI.

Businesses need to remind their employees of the responsibilities they have and train them to be vigilant with technology. In this way, a culture of resilience can ultimately be built in the organization, based on shared standards of trust and ethics.

At SAS Innovate, this cultural aspect is central. The company has set itself six principles for responsible innovation. In addition, there is intensive collaboration with customers, partners and scientists to raise awareness of the need for ethical, fair and sustainable AI.

“You need to know where your technology can go and how to use it to make sure it doesn’t end up causing harm to people. Our vision is a world where data helps people succeed. If we can help those most vulnerable by co-innovating alongside responsible innovators, we can be sure the rest will do well too,” concludes Townsend.

This is an editorial in association with SAS. Learn more about the company’s solutions here.