绿巨人视频

On Wednesday, April 27, 2022, the Office of 绿巨人视频鈥檚 President Celeste M. Schenck hosted the tenth and final event in its Presidential Lecture Series: a talk by Jon Iwata, Executive Fellow at Yale School of Management and Executive Director of the Data & Trust Alliance. The Presidential Lecture Series, titled 鈥淭echnology and the Human Future,鈥� invites speakers to participate in live online events, so they might engage with both theory and practice in responding to the question of how technology will continue affecting our lives beyond the Covid-19 pandemic. Wright presented on the topic of 鈥淒ata, Artificial Intelligence and Our Responsibilities.鈥澛�

Prior to Yale, Jon Iwata led a 35-year career at IBM, reporting to three different CEOs during a period of significant transformation in the IT industry. As Senior Vice President of Marketing and Communications, he led the company鈥檚 global marketing, communications and corporate citizenship organization for a decade. Following his retirement in 2018, Iwata helped found the Data & Trust Alliance in 2020, a consortium of leading companies working to develop and promote responsible data and AI practices and policies.

Iwata began his talk by addressing some misconceptions about artificial intelligence, explaining that AI is not designed to replace humans, but rather to augment our ability to perform certain tasks. 鈥淎I is just another tool in a long history of tools civilization has created,鈥� he explained. He categorized AI as similar to other high-powered technologies that extend human abilities, such as the Large Hadron Collider and James Webb Space Telescope. But whereas these tools improve human senses, AI is primarily an extension of human decision-making abilities.

Databases exist today that cover a wide range of elements that impact human decision-making, from the products we buy to the choices we make about our healthcare provision. Iwata explored how data on geolocation, genetics and even sleep patterns are affecting corporate strategy worldwide. 鈥淭he promise of AI is that the accuracy of decisions will increase, and the cost of predictions will drop to almost zero,鈥� he said, comparing the growing ubiquity of AI-powered decision-making to the way in which the internet reduced transaction costs to be almost negligible.

But Iwata also warned of potential risks of using AI in decision-making, such as bias affecting the algorithm training process or that a lack of transparency in the way certain algorithms learn will make monitoring impossible. Iwata鈥檚 work with the Data & Trust Alliance is geared toward preventing such problems. The consortium focuses on practical implementation, building good business practices that can easily be adopted by a wide variety of business professionals. 鈥淎doption is our one KPI,鈥� he explained.

Iwata then presented the consortium鈥檚 first-ever project 鈥� on mitigating unfair bias in workforce decisions. AI use in HR and hiring practices is rapidly growing. The project offers ways for companies to evaluate the safeguarding practices of their vendors. It involves a detailed evaluation process, including questionnaires, scoring mechanisms and HR and procurement training packages. Walmart, the US鈥檚 second-biggest employer, is among the companies now using the package.

Iwata concluded with some advice for students. 鈥淵ou are citizens today and leaders tomorrow,鈥� he said. 鈥淵ou will use data and AI as both consumers and professionals.鈥� He laid out five questions he advised students ask in order to deploy AI responsibly in the future. 1) Is the model necessary and does it offer more risk than value? 2) Who built and reviewed the model? 3) Was bias considered at the design stage of the model? 4) Can you effectively monitor how the model changes over time? 5) How will the model approach the new literacy necessary to interpret its terms and vocabulary?

After his lecture, Iwata took questions from the online audience. You can watch his whole talk and Q&A in the video below.

The entire Presidential Lecture Series is now available to watch online.