We have been seeing the potential of AI as big techs have been pouring money into their own and making it more advanced, trying to up the other. As good as that can be, many will find the opportunity to use it wrong, which can affect US elections.
AI Contributing to Misinformation
The generative power of AI could prove to be harmful, seeing as it still can't detect real news from fabricated ones. This could be a problem for the US presidential elections this 2024, given that some candidates might be subjected to false information that damages their reputation.
For instance, AI from ElevenLabs is now capable of cloning voices, while OpenAI's DALL-E 2, Stable Diffusion, and Midjourney, as mentioned in Interesting Engineering, can create realistic photos from samples, which can easily be altered with just a few sample words.
Fake news has always been a tool used by people on all sides to either garner support for their candidate or tear the competition down. AI is yet another tool that can help with a smear campaign, which justifies the concerns of experts.
A professor at New York University and expert in AI, Gary Marcus expressed that people should be scared. He added that it was hard to see how AI-generated misinformation will not be a "major force" in the 2024 election.
A fellow at the Brookings Institution, Chris Meserole warns that voice clones can also be used to influence the elections. Fake audio clips can be released at crucial moments in the election wherein a candidate is saying something disreputable, effectively swaying others against them.
In some services that provide AI-generating images, you can simply upload a photo and instruct it to put that subject in a given context. It can be used to create fake images wherein a candidate is doing something appalling which could take votes away from them.
There's also the issue of people believing everything they see on the internet. While others try to verify the information, some will simply take it as truth. As advanced as AI is now, fake images, audio, and videos can appear very realistic and convincing.
Read Also: OpenAI Launches AI-Written Text Detection Tool: AI Text Classifier
Detecting AI-Generated Content
As AI technology progresses, AI tools to detect AI-generated content have also been developed. This is to prevent instances such as passing AI content off as original or real audio, video, or image.
While images and audio are harder to detect, there are various sites where you can copy and paste the content, and the service will analyze it for you. Although it's still not as accurate as others would hope, it still helps.
There is such a service called ZeroGPT. Users can enter the text in the empty box and click "Detect Text." There are many results that could appear such as "Your text is Human Written," "Your text is AI/GPT Generated," "Most of Your text is AI/GPT Generated," and more.
Of course, technology has also given us the tools to detect fake audio, videos, and images. However, it is not as simple as copy-pasting text in a box and let a website do the work. Some might need an experienced person to determine whether it's real or fake.
Related: 5 Risks of AI-Generated Content
