Apple can only resist AI transparency for so long

It's only a matter of time until the demands for ethical accountability in AI become unavoidable.

Apple, AI, Gen AI, Mac, IPhone

Apple’s 2024 Proxy Statement for its annual shareholder’s meeting reveals several little-known initiatives it's been following to improve artificial intelligence (AI) across its product range, and the company seems to go a little further than others to attend to the details.

The company shared insights into how it thinks in response to a shareholder proposal supported by two corporate investors — Legal & General Investment Management (one of the top 25 Apple stockholders) and Abrdn.

The investors want Apple to publish a transparency report on the company's AI usage and ethical frameworks. Apple had tried to exclude the resolution, but the US Securities and Exchange Commission (SEC) dismissed that attempt.

Should AI roam free?

The central argument of the shareholder proposal is respectable. No one should be blind to the deeply transformative impact AI is about to have on every society and every profession. The legal profession, for example, is preparing for one of the most deeply disruptive impacts it has ever experienced from tech.

The drive to force companies in the business of building AI and generative AI tools and platforms to adhere to ethical standards will only intensify as populations wake up to the significance of mass deployment. In the end, Apple’s perceived slowness in widening AI deployment in contrast to some competitors might reflect that the company sees things the same way.

“We believe it is important to be deliberate and thoughtful in the development and deployment of artificial intelligence, and that companies think through the consequences of new technology before releasing it,” Apple said in response to the shareholder proposals.

Apple thinks about it seriously

Apple also shared a couple of examples to back up its argument:

  • “Apple is engaged in an ongoing process to make Siri a more inclusive and accessible feature, including by engaging socio‐linguist experts to improve speech recognition accuracy rates for users of different ethnic/racial, gender, and geographic backgrounds and working in partnership with Black and African American Vernacular English‐speaking volunteers,” the company said.
  • Apple also says that when it was developing Face ID, the company was very conscious that facial recognition algorithms had until then not been terribly accurate in recognizing diverse demographic groups. To tackle this, its teams worked with volunteers from around the world to include a representative group of people accounting for gender, age, ethnicity, disability, and other factors when building FaceID.

Both those efforts seem significant given the 2018 warning from Apple’s former Senior Director of AI and Machine Learning, Carlos Guestrin, that poorly implemented AI could exacerbate societal prejudice.

“It’s not enough just to think about the data that we use but also how that data reflects our culture and values that we aspire to,” Guestrin said at that time. “There is now a trend not just to thinking about the business metrics, but the social impacts machine learning will have.”

More AI transparency is inevitable

Guestrin’s comments suggest there’s something to be said for forcing companies to reveal their ethical approach to the use and deployment of AI technology. The idea that decisions of huge consequential impact could be made by machines on the basis of principles and ethics that are not transparently shared or open to examination cannot stand.  

The shareholder proposal warns against AI being used to undermine privacy and generate fake media content, and even notes the impact machine intelligence is already having on jobs and workplaces.

However, as Apple points out, the scope described in the original proposal is very broad and could in principle force the company to share confidential R&D information.

While the resolution might be too broad in scope, populated as it seems to be by a  potpourri of current AI concerns, the inconvenient truth is that the time within which these technologies can be deployed without oversight is limited. In the end, the winners in the ongoing game might — or might not — match Apple’s steadfast dedication to considering the impact it could have before unleashing it.

Please follow me on Mastodon, or join me in the AppleHolic’s bar & grill and Apple Discussions groups on MeWe.

Copyright © 2024 IDG Communications, Inc.

It’s time to break the ChatGPT habit