Covering Disruptive Technology Powering Business in The Digital Age

By: Martin Dale D. Bolima, Technology Journalist

Maybe the notion that Artificial Intelligence (AI) could go rogue someday is not that far-fetched after all. Or at least the idea that it can pose a danger to society at large.

OpenAI’s Sam Altman talked about that in his exclusive interview with ABC News. The one-on-one sitdown aired on Friday in the U.S. Altman, the CEO of OpenAI spoke at length about the recent launch of ChatGPT-4. He also covered other topics related to AI, like machines potentially replacing human jobs.

Altman admitted to that possibility and worries it will happen quickly. But he believes humans will adapt, as they have done over the years by adapting “wonderfully to major technological shifts.” To adapt, said Altman, people must consider ChatGPT as a tool to augment work, not a replacement for people.

The Persistent Problem of Misinformation

The OpenAI CEO also discussed the possibility of people using AI for potentially dangerous applications. The thought of AI being used for bad things, according to Altman, often causes him to lose sleep. And his biggest concern at the moment is GPT and AI in general being used for disinformation.

“I’m particularly worried that these models could be used for large-scale disinformation,” Altman admitted. “Now that they’re getting better at writing computer code, [they] could be used for offensive cyberattacks.”

OpenAI touts ChatGPT-4 as a massive improvement from ChatGPT-3. However, only a small set of users can access it at the moment. Among these select early users are a group of visually impaired users who are part of ChatGPT-4’s beta testing.

Despite the purported improvements to ChatGPT-4, Altman admits it still spews misinformation. It is, according to Altman, a consistent problem with AI language models like ChatGPT, which routinely gives inaccurate information. Altman attributes this problem to what he terms “hallucinations problem.”

“The thing that I try to caution people the most is what we call the ‘hallucinations problem,'” Altman pointed out. “The model will confidently state things as if they were facts that are entirely made up.”

The Human Takeover Problem

While he made these damning admissions, Altman clarified, however, that the greater threat is not AI itself but rather humans themselves. According to Altman, he worries about who takes control of AI, not about AI growing sentient and plotting a worldwide takeover.

“It [AI] waits for someone to give it input. This is a tool that is very much in human control,” explained Altman. “[But] there will be other people who don’t put some of the safety limits that we put on. Society, I think, has a limited amount of time to figure out how to react to that, how to regulate that, how to handle it.”

To that point, Altman referenced something Russian President Vladimir Putin said back in 2017. Putin boldly declared whoever controls AI would likely rule the world as well. And Altman agrees with it in a sense.

“That’s a chilling statement for sure,” Altman admitted. “What I hope, instead, is that we successively develop more and more powerful systems that we can all use in different ways that integrate it into our daily lives, into the economy, and become an amplifier of human will.”

Altman, however, recognizes the human control conundrum. And that he says can be a problem down the line.

“A thing that I do worry about is … we’re not going to be the only creator of this technology,” Altman pointed out. “There will be other people who don’t put some of the safety limits that we put on it.”

Hard to Control Everything . . . and Everyone

For its part, OpenAI put in place a team of policymakers tasked to filter the information that goes into ChatGPT.  This same team also manages what ChatGPT can share with users to prevent it from providing objectionable or illicit content.

“[We’re] talking to various policy and safety experts, getting audits of the system to try to address these issues and put something out that we think is safe and good,” Altman emphasized. “And again, we won’t get it perfect the first time, but it’s so important to learn the lessons and find the edges while the stakes are relatively low.”

But, as Altman pointed out, OpenAI is not the only one developing AI. And as much as the company tries to build a responsible model, it cannot control what other developers are doing. Neither can it control how people are using AI.

Those thoughts worry Altman very much. And given the advances in AI, maybe everyone should worry, too.

(0)(0)