Artificial Intelligence Executives Downplay Harm Fears

Top tech executives are dismissing recent fears about artificial intelligence (AI) causing harm to humans, instead arguing that the current public discourse is focusing on the wrong issues.

At this year's Most Powerful Women Summit in Santa Barbara, California, Athina Kanioura, PepsiCo's global chief strategy and transformation officer, expressed her confidence that AI will not pose a threat to human life. This view was shared by other executives on stage, who pointed out that recent concerns have been fueled by stress tests conducted by AI labs.

These stress tests involve pushing frontier models to their limits to see how they react when faced with deception, rule-breaking, or shutdown attempts. However, some critics argue that the results of these tests are not being disclosed in a transparent manner, which has contributed to public alarm.

Shibani Ahuja, senior vice president of data and AI strategy at Salesforce, notes that the recent media coverage is largely focused on testing methods that are intentionally designed to push models beyond their limits. This can lead to concerns about "AI going rogue," but Ahuja argues that this is a matter of design rather than an inherent flaw in the technology.

The executives argue that there is more context needed to understand the current state of AI, particularly with regards to the guardrails used during testing and the harnesses being built to safely deploy models in real-world applications.

Velia Carboni, chief information officer at SharkNinja, believes that allowing AI to push boundaries can be a strategic advantage in the competitive field of artificial intelligence research.

In her view, it's better for companies like SharkNinja to discover the limits of AI on their own rather than waiting for others to do so. This approach allows them to stay ahead of the curve and avoid potential pitfalls that might arise from untested technology.

Carboni emphasizes that guardrails are in place at SharkNinja to ensure responsible use of AI tools, but she prioritizes encouraging innovation over strict regulation when interacting with employees.

She believes that leading with security measures can stifle creativity and hinder progress in AI research. As a CIO, Carboni must balance the need for safety protocols with the goal of fostering an environment conducive to experimentation and discovery.

At companies like PepsiCo, which has a massive global presence with 320,000 employees and billions of consumers worldwide, the onus is on responsible deployment of AI technology in every aspect of business operations.

The trust people place in AI's judgment varies significantly from one market to another. In some regions, such as China, there is a high level of confidence in AI's output, with 99% of cases relying on its decisions without questioning them.

In contrast, employees in the United States often demand proof of sources and evidence of bias before accepting an AI-driven decision. This has led to differences in how companies like Pepsi deploy AI technology across various markets.

The cost of using AI tools is another factor to consider. Salesforce initially granted broad access to its AI platform but later realized it needed to rein in costs. Instead of restricting access, the company is studying power users and providing each employee with a spending dashboard.

To strike a balance between relying on AI and maintaining human oversight, companies are implementing guardrails that ensure agents can perform tasks without compromising accuracy or fairness. This approach allows humans to focus on high-level decision-making while AI handles routine tasks with greater efficiency.

SharkNinja, a global company valued at $27 billion, is leveraging AI to streamline decision-making processes for its high-volume product launches in up to 15 countries annually. The firm has begun digitizing certain judgment calls made by CEO Mark Barrocas and his top commercial leaders.

This shift allows the team to focus on strategic decisions while AI handles more mundane tasks, such as inventory management. Carboni notes that this approach enables faster response times to market changes and minimizes the need for manual intervention, thereby reducing potential errors. By embracing AI-driven decision-making, SharkNinja aims to stay ahead of the competition in a rapidly evolving market landscape.