Fact Check: How to spot AI images?

Correction: Xi Jinping is China's president and not the premier as said in the video.

Images that look deceptively real but were created with the help of artificial intelligence are flooding the Internet. Here are 6 tips on how we can tell real from fake.

#factchecker #ai #fakenews

Follow DW on social media:

[dwnews]

Correction: Xi Jinping is China's president and not the premier as said in the video.

[Tom-tgjl]

These little errors won’t last long, stuffs getting better everyday

[Jon-btzv]

Nice fact check. But I agree it’s becoming more difficult to detect.

[sommmeguy]

So how about instead of putting yet another burden on the average person, we use the power of government to force the AI companies to invent an authentication system before they are allowed to operate in our communities?

[arrow]

One way is noticing how perfect and flawless the complexion on people's faces are. No moles, pimples and warts or even stubble and the skin tone is a perfect uniform color even in complex lighting conditions.

[NinjaRunningWild]

AI vs people’s critical thinking skills, _AI wins._

[noaht]

This video will be relevant for a few months... Then the AI will correct all of its flaws. 🤷‍♂️

[robertalker]

Was a time when pictures never lied. We should mourn the passing of that time.

[okman]

Looks like AI is gonna make better memes than human

[Icantkeepout]

I don't like AI because art is from a the creators eye, heart and interpretation of their vision and not and not computer manipulation.

[SimonGreene]

Quality journalistic work. Good Video.

[matsf]

So in 20 years we will shut down internet because we cant trust anything but IRL meetings?

[TCN]

Some people are really spooked by these technological advances. I would argue, we should be spooked by the unintelligent people who are still taking out-of-context images at face value.
1. Out-of-context sound bytes have long been used to trick, cheat, compromise etc. What's so different with images ?
2. Doctored images are easier to make than AI-generated images and just as insidious and effective - if not more. So nothing new under the sun.
3. Police, administration and journalists should be trained to know of the technological capabilities and apply - or reserve judgment accordingly. Forewarned is forearmed.

[dandiesel]

Good luck with that because within the next year, its going to be impossible to tell the difference between AI and whats real.

[bartmannn]

Very concerning that these tools, that are obviously designed for spreading misinformation aren't legally classified as such, the same as weapons are. If someone gave out weapons and ammunition in the public, for free, this person would obviously be held legally accountable. AI developers, on the other hand, take no responsibility whatsoever - no one puts them in jail.

[kosta]

Just wait two years and I'm sure that most of those dead giveaways will be perfected by AI.

[matthewdugas]

The united states would do better with an ai chat bot running the supreme court. It could be anywhere as corupt as some that are in there now. If all government elected officials and appointed justice had to pass phycological evaluations and there wwere stricter penalties and strickter code of ethics. We would have only a hand full of people in our goverment

[hohohaha]

What are the chances someone plugs this video into an AI to make better images with an AI?

[Peleski]

I get a lot of really amazing plants and flowers in my SMedia feeds. Because of genetic engineering and hybridisation, it's hard to know if they're AI or not. People do breed some amazing varieties nowdays.

[UkrainianBullwinkle]

Biggest thing the world can do is try and shut down ai websites and stop selling computers extremely powerful and down grade jusstta bit