Advertisement

Sam Altman's election gameplan shows he's desperate to avoid repeating Mark Zuckerberg's mistakes

Sam Altman
OpenAI chief Sam Altman will be keen to avoid Facebook's election mistakes.Kent Nishimura
  • Sam Altman has a huge task in Davos: convince leaders he won't repeat Facebook's election mistakes.

  • OpenAI has outlined its plan to prevent tools like ChatGPT from being used to interfere with elections.

  • The game plan includes a "proactive approach" to deep fakes and a ban on political campaigning apps.

Sam Altman will be on a charm offensive in Davos.

As the ChatGPT boss visits the Swiss Alps this week for the World Economic Forum, you can bet he'll field lofty questions about the future of AI from the usual parade of elites who attend the annual retreat.

The fretful among them will have a much more pressing question to ask: what's his plan to avoid Mark Zuckerberg election mistakes?

As the US and other nations gear up for elections this year, the political class will have unfavorable recollections of the role Facebook played in hurting the democratic process over the last decade.

Under Zuckerberg's watch, Facebook facilitated the spread of misinformation, became the target of bad actors from Russia abusing its service, and allowed the data of millions of users to be harvested by a company that sought to swing voter intentions with micro-targeted ads.

In the wake of the 2016 election, Zuckerberg dismissed the suggestion that Facebook influenced the vote as a "pretty crazy idea." He said voters had made "decisions based on their lived experience" and that fake news constituted a small portion of content on the platform.

Two years later, in 2018, he appeared before Congress and conceded the company "didn't do enough" to, among other things, prevent foreign interference in elections.

Concerns that AI could have an impact on elections are not unwarranted. Since the launch of ChatGPT there have been plenty of examples of examples of generative AI "hallucinating" – or creating inaccurate information.

Political leaders will be desperately hoping that Altman doesn't allow ChatGPT to cause the level of havoc Facebook previously has come election season.

Good thing Altman has turned up in Davos with a plan.

Mark Zuckerberg testifying in the Hart Senate Office Building in April 2010 over the Cambridge Analytica scandal. He is sitting in front of a plaque that has his name on it.
Meta's Mark Zuckerberg was grilled before congress over the company's role in spreading misinformation.Chip Somodevilla/Getty Images.

On Monday, OpenAI released a blog that detailed its plans for approaching worldwide elections in 2024.

"Protecting the integrity of elections requires collaboration from every corner of the democratic process, and we want to make sure our technology is not used in a way that could undermine this process," the blog's opening statement said.

There are a few different steps being taken by OpenAI's teams – including the safety systems, threat intelligence, legal, engineering, and policy departments – to ensure that.

The first, it says, is a proactive approach to preventing its tools from being used to create deep fakes, as well as "scaled influence operations, or chatbots impersonating candidates."

For instance, OpenAI says DALL-E, a text-to-image model, "has guardrails to decline requests that ask for image generation of real people, including candidates."

The company is also barring people from using its tools to build applications for "political campaigning and lobbying," until it figures out how effective they are in persuading people.

Transparency is a big talking point for OpenAI too.

In a world becoming increasingly filled with AI-generated content, it can be harder to determine what is and isn't AI-generated. For that reason, tools are being introduced for voters to distinguish between them.

ChatGPT, meanwhile, will be "increasingly integrating with existing sources of information," such as real-time news reporting, so that users can see the attributions of information they're getting from the chatbot.

How much of this will make a difference is to be determined. AI is already out the box and in use for campaigning. Republicans used an AI-generated ad last year to depict a bleak, imaginary future for the US if Joe Biden was re-elected.

Altman will hope these safeguards will be enough. In a year where over 64 countries go to the polls, it could hardly be more important that they do.

Read the original article on Business Insider