photograph: Jakub Polsicki
The use of artificial intelligence (AI) in political campaigns needs stronger regulation to prevent voter manipulation, warns an Oakland law lecturer.
The People’s Party confirmed this week that it used AI to generate images for the election campaign.
On Thursday, Privacy Commissioner Michael Webster released a list of expectations that should apply to the private and public sectors when using generative AI.
But Nikki Chamberlain, a senior law lecturer at the University of Auckland and an AI law expert, told RNZ. morning report I needed more.
“I think we need to be transparent, and we need to know where the images and videos that are created as part of the marketing of political campaigns come from. Because I think it will lead to things,” she said. “But I also think we need a law to actually mandate this.”
Chamberlain – also one of the editors of the latest 3rd edition. New Zealand Privacy Law Textbook – The 2020 Privacy Act has provisions governing how people’s personal information is used, shared and analyzed, but said it had some weaknesses.
“This is not the highest standard, like the EU General Data Protection Regulation. There is a need for stronger enforcement of the Privacy Act as the maximum penalty under the Privacy Act is NZ$10,000.”
University of Auckland Law Lecturer Nicky Chamberlain
photograph: Supplied
He said New Zealand also needs more specific regulation on the use of AI “to ensure voters are not manipulated and to ensure transparency”.
He said there are two main types of AI used in political campaigns: machine algorithms, such as those seen in the Cambridge Analytica scandal, and generative AI, such as ChatGPT and Midjourney.
The former was of concern because it could be used to target users with political ads to manipulate them into voting for a particular political party or candidate.
“There was an application and a quiz on Facebook and people took part in it … their information and their friends’ information was sucked into it and it was used by Cambridge Analytica to basically determine voter preferences.” said Chamberlain.
“For example, if you see that someone keeps clicking on crime-related links, or is particularly interested in crime-related issues, you will take advantage of it. are scared, so they will target ads around the theme of fear.” It is also important to strengthen police presence and crack down on crime.”
Generative AI, on the other hand, takes a collection of data, such as a sequence of words or images, and uses it to create new data that closely resembles the type of input data. This means that a person can request a particular type of image or document, and the AI will generate what it expects someone to see based on previous examples the software has access to.
It’s the sort of technology behind programs like ChatGPT for quickly creating text documents and Midjourney for creating images, which National used in its advertising campaigns.
Chamberlain said ostensibly the only problem would be stock image actors losing their jobs, but it could be exploited by things like video AI like deepfakes.
“The worse part of the problem is that if it’s unchecked, it creates the image that people are saying things they aren’t saying and voters don’t know it.” [that it’s not real].
“It’s easy to get sucked into a story that may not be accurate.”
He said campaign materials using AI should make that clear to ensure transparency, and that laws need to be updated to prevent further abuse of the burgeoning technology.
“The biggest thing I would say is to encourage political parties using AI to state in their ads that they actually use AI. You can tell where it came from and it’s not real.”
She also left a message for voters to “check your sources and refer to multiple sources to make sure the information is accurate.”
