How convincing are AI-generated deepfakes - and should we be worried?
When you can fabricate a video of anyone saying anything, truth itself becomes negotiable.
The answer
Deepfake technology, which uses deep learning to superimpose faces and synthesize speech, advanced dramatically between 2017 and 2023. By 2023, AI-generated faces were indistinguishable from real ones in studies where human evaluators scored only 48.2% accuracy - worse than a coin flip. The technology has been used for fraud (a 2023 deepfake video call tricked a company into transferring $25 million), political disinformation, and non-consensual intimate imagery. Detection tools exist but face an asymmetric battle: generation models improve faster than detection methods, creating what researchers call a 'deepfake arms race.'
Source: PNAS
- deepfakes
- misinformation
- ai ethics
- computer vision
More from AI & Machine Learning
- What radical question did Alan Turing ask in 1950 that launched the entire field of artificial intelligence?
- Can a machine fool you into thinking it's human - and does that mean it's truly intelligent?
- What happened at a small Dartmouth workshop in 1956 that gave artificial intelligence its name?
- How did a 1958 invention called the Perceptron convince people that machines could learn like brains?
- Could a 1966 chatbot really make people believe they were talking to a therapist?
- What was the first robot that could 'reason' about its own actions?