Scientists say 偽の 直面するs created by AI look MORE real than human 直面するs - so, can you tell which of these are actual people?

  • AI 'hyperrealism' means 偽の 直面するs are perceived as more real than human ones?
  • READ MORE:??偽の 直面するs created by AI appear more 信頼できる, 熟考する/考慮する 明らかにする/漏らすs

人工的な 知能 (AI) is now so sophisticated?that we can't tell the difference between 偽の 直面するs and snaps of real people, a new 熟考する/考慮する 警告するs.?

In 実験s with US 国民s, more people thought AI-生成するd 直面するs were human than the 直面するs of real people.

専門家s are 関心d that 'hyper-現実主義の' imagery could be 燃料ing misinformation and 身元 窃盗 online by creating authentic-looking profiles of people.?

In the 熟考する/考慮する, the 研究員s compared five AI 直面するs with five human 直面するs.?

So, can you tell which of these people are real? Scroll 負かす/撃墜する for the answers.

In the study, the researchers compared five AI faces with five human faces. So, can you tell which of these people are real? Scroll down for the answers

In the 熟考する/考慮する, the 研究員s compared five AI 直面するs with five human 直面するs. So, can you tell which of these people are real? Scroll 負かす/撃墜する for the answers

READ MORE AI 直面するs 'look more 信頼できる'

Researchers are calling for safeguards to prevent 'deepfakes' from circulating online

研究員s are calling for 保護(する)/緊急輸入制限s to 妨げる 'deepfakes' from 広まる online?

?

宣伝

The new 研究 was led by?Dr Amy Dawel, a?cognitive and 臨床の psychologist at?Australian 国家の University (ANU) in Canberra.?

AI 直面するs fool people into thinking they're real because they mimic typically human ideals of what we look like, によれば the authors.?

'It turns out that there are still physical differences between AI and human 直面するs, but people tend to misinterpret them,' said?Dr Dawel.?

'For example, white AI 直面するs tend to be more in-割合 and people mistake this as a 調印する of humanness.

'However, we can't rely on these physical cues for long.?

'AI 科学(工学)技術 is 前進するing so quickly that the differences between AI and human 直面するs will probably disappear soon.'

For the 熟考する/考慮する,?Dr Dawel and 同僚s 新採用するd 124 居住(者)s in the US, all of whom were white and 老年の between 18 to 50 years.?

They were shown 100 real 直面するs and 100 AI 直面するs that were 生成するd using?StyleGAN2, an AI 道具 created by US 会社/堅い?Nvidia.

After deciding whether a 直面する was AI or human, 関係者s 率d their 信用/信任 on each 裁判,公判 from 0 (not at all) to 100 (完全に).?

White AI faces are, remarkably, judged as human more often than pictures of actual humans. Perceptual qualities of faces that contribute to this 'hyperrealism' phenomenon include facial proportions, familiarity and memorability

White AI 直面するs are, remarkably, 裁判官d as human more often than pictures of actual humans. Perceptual 質s of 直面するs that 与える/捧げる to this 'hyperrealism' 現象 含む facial 割合s, familiarity and memorability?

The answers?

  1. AI 女性(の)
  2. Human male
  3. Human male
  4. AI 女性(の)?
  5. AI male
  6. Human male
  7. AI male
  8. AI male
  9. Human 女性(の)
  10. Human male?

宣伝

Worryingly, four out of five of the 直面するs 裁判官d as human most often by the 関係者s were 現実に AI.

一方/合間, four out of five of the 直面するs 裁判官d as AI most often were 現実に human.

'White (but not 非,不,無-white) AI 直面するs are, remarkably, 裁判官d as human more often than pictures of actual humans,' the authors say.

'We pinpoint the perceptual 質s of 直面するs that 与える/捧げる to this hyperrealism 現象, 含むing facial 割合s, familiarity, and memorability.'?

Worryingly, people who thought that the AI 直面するs were real most often were the most 確信して that their 裁判/判断s were 訂正する.

'This means people who are mistaking AI ペテン師s for real people don't know they are 存在 tricked,' said co-author Elizabeth Miller at ANU.?

Ironically, the 職業 of 正確に telling the difference between AI 直面するs and real 直面するs may 落ちる to a machine, the authors 示唆する.?

'Given that humans can no longer (悪事,秘密などを)発見する AI 直面するs, society needs 道具s that can 正確に identify AI ペテン師s,' Dr Dawel said.

'Educating people about the perceived realism of AI 直面するs could help make the public 適切な 懐疑的な about the images they're seeing online.'

The authors are 関心d about a racial bias, because AI algorithms are trained 不均衡な on white 直面するs and so may appear more 現実主義の.?

AI depictions of 黒人/ボイコット 直面するs may be more ina ccurate, although this would mean they're more easily discernable from photos of real 黒人/ボイコット people.?

'If white AI 直面するs are 終始一貫して perceived as more 現実主義の, this 科学(工学)技術 could have serious 関わりあい/含蓄s for people of colour by 最終的に 増強するing racial biases online,' said?Dr Dawel.?

'This problem is already 明らかな in 現在の AI 科学(工学)技術s that are 存在 used to create professional-looking headshots.?

'When used for people of colour, the AI is altering their 肌 and 注目する,もくろむ colour to those of white people.'

The 熟考する/考慮する has been published today in the 定期刊行物?Psychological Science.??

Instagram photographer 収容する/認めるs his portraits are all AI-生成するd after 伸び(る)ing thousands of 信奉者s

An Instagram photographer who raked in thousands of 信奉者s thanks to his 素晴らしい portraits has 認める that they are 現実に 偽の.

Jos Avery, from the US, even gave each 人物/姿/数字 a 指名する, geotag and moving story that he 追加するd in the image captions.

The imagery was created by Midjourney - a ソフトウェア that uses 人工的な 知能 (AI) to 生成する images - and touched up on Photoshop.

The photorealistic, 黒人/ボイコット-and-white images saw his Instagram account 伸び(る) nearly 29,000 信奉者s since he first 地位,任命するd in October 2022.??

Read more