Experts say most Americans believe AI companies prioritize profits over safety, and recent missteps have reinforced that perception.
As advances in AI are touted and promoted as being of great value to business and people's lives, the industry has been given a bad rap. The latest example is OpenAI's attempt last week to surprise the world with improvements to ChatGPT's voice capabilities.
Actress Scarlett Johansson has accused the company of copying her voice from “Her,” a Spike Jonze romance about a lonely writer who falls in love with her AI companion. OpenAI has denied the accusations. washington post The company this week cited records showing it did not copy Johansson's voice.
The controversy comes the same week that Jan Reicke, who led OpenAI's safety team, left the company, saying on X (formerly Twitter) that the company's “safety culture and process had taken a back seat to flashy products.”
OpenAI disbanded Reich's Super Alignment team after his retirement. Other companies that have reorganized or scaled back their AI safety efforts include Amazon, Google, Meta, Microsoft and X, amid growing criticism that profits override ethics.
“There's a rampant AI ethics bluff,” said Kashyap Komphela, CEO of consulting firm RPA2AI Research. “Lots of companies are making big pronouncements about responsible AI principles, but most of it is just play-acting, a box-ticking exercise.”
Concerns about AI
The AI industry's actions have tarnished its image in the public eye. Researchers at the Markkula Center for Applied Ethics at Santa Clara University in California conducted a poll of 3,000 American adults last November, finding that 68% were “very concerned” or “somewhat concerned” about the impact of AI on humanity.
The poll found that most Americans distrust AI companies, with 55% agreeing that ethics aren't considered when developing AI. This lack of trust led 86% of Americans surveyed to believe AI companies should be regulated, and 83% to agree that the government should develop clearer AI regulations.
trust [in AI] That's probably the lowest level I've seen in the several years I've been paying attention.
Irina ReikDirector of the Internet Ethics Program at Mark Kula
“Trust is probably at the lowest level I've seen in the last few years I've been paying attention,” said Irina Reik, director of Markkula's Internet Ethics Program. She says, “People are just afraid of their motives, their impact on society, and whether they're really going to carefully manage the tools they're developing.”
Ann Skeet, senior director of leadership ethics at Markkula, said this credibility gap could impact investors and sales for many companies evaluating and purchasing AI technology. Since the start of the AI boom in 2022, starting with the launch of OpenAI's ChatGPT, businesses have been plagued by data privacy and security concerns.
“Reputational damage can lead to loss of shareholder value,” Skiat said. “Any significant failure could result in loss of reputation in the marketplace.”
The government is also scrutinizing the AI industry, along with the public and companies. California, Colorado, Illinois, and Texas have introduced or passed laws to protect the public from AI abuse. In March, the European Parliament approved the European Union AI Act, which provides a regulatory framework for the operation of AI systems.
“It shows that the public and regulators are tired of relying on these internal control efforts to keep us all safe,” Reich said. “The public feels that these efforts are not enough and we need some kind of legislation.”
Despite the setbacks, the AI industry can change the perception that it cannot be trusted. Markkula's poll showed that a significant portion of respondents remain optimistic, with 48% agreeing that AI is currently having a positive impact on their lives and 45% still saying they are “very excited” or “somewhat excited” about AI's potential.
And some experts are optimistic that the industry will bounce back and act responsibly.
“I don't think it's impossible in principle,” says Michael Bennett, director of the Responsible AI Program at Northeastern University. “Now, would I bet my retirement money on it? I wouldn't, but I do think it's possible.”
That's certainly a possibility, but only after dramatic changes occur, experts say.
“For these companies to build trust with consumers, they need to demonstrate that they are serious about developing reliable technology,” Skeete said.
Antone Gonsalves is Editor-at-Large for TechTarget Editorial, reporting on industry trends important to enterprise technology buyers. He has been in technology journalism for his 25 years and is based in San Francisco.
