The dangers of AI to cryptozoology

[Shazzoir]

Morning all,

In the past six months, you will all no doubt have become aware of the technology commonly known as 'AI' or 'Artificial Intelligence'.

Already, AI and machine learning-enabled technologies are used in medicine, transportation, robotics, science, education, the military, surveillance, finance and its regulation, agriculture, entertainment, retail, customer service, and manufacturing, just to name a few.

AI isn't new, especially to the various military agencies across the world, but it's pervasive, and in use in many common everyday ways.
For example, if you're Googling something, you're making use of AI.
If you're using a smart phone with autocorrect, or Siri etc., you're making use of AI.
If you're watching movies, you're watching CGI and other AI.
Same goes for the 'smart home' management devices and home alarms and cameras like Ring, Eufy and Alexa. AI is used in the medical field to improve the accuracy of programs used for detecting health conditions. It's technology is used in smart TVs and movie, TV and music programs like Netflix or Spotify to monitor a user's habits and make recommendations based on their recent activity. Facebook uses AI to do the same, as does other social media; in short, it's all around us and making impacts on our lives.

I can live with all this, though I don't personally have most of those things mentioned above, but as an artist, the thing that pisses me off the most, is how AI is being used to create art and images, many of which find their way into the fertile bullshit-fed algorithms of Facebook. In past months, I've seen AI-generated images galore all over FB, Imgur and other pages, and while this would be fine if the posters stated it was AI, I rarely see this occurring. No attribution is given, as anyone can make use of the multitude of AI pages available, many of which are free to use like Craiyon, which is fairly basic. Midjourney however, is a professional-level program that makes stunning images, and this is where the problem lies.

AI has already begun to make it virtually impossible to tell if an image is real or not.
Some people are already using it to generate images like this one below, but are they doing so as a beneficial tool, for education and awareness? Or are they doing it to 'fool' people, make a mockery of the study and interest in cryptozoology?

AI pic.png

How long before we start seeing 'fake' images here on AYR and other crypto pages? I mean, we all know there's a certain element of the public who will bullshit and lie, and make up fake evidence to gain attention or some kind of social kudos or monetary gain?(as in YouTube channels that are monetized per view)

AI as the name suggests, is intelligent in that it 'trains' and improves itself via machine learning. The day will come when I reckon we'll be bombarded by AI generated images and video of Yowies, Thylacines, Thylacoleo, extinct megafauna... the list is endless.

How as a repository of GENUINE reports and media, do we here at AYR propose to manage this?
We could look at two options where one is continue as we are, and have the Admins and Mods delete posts that are suspected AI fakes, OR
We make use of this AI to our own benefit, and create a section in the Forum for people who WANT to experiment with AI, so they can show off their work and the images they can generate. It would be deemed 'AI' work by default if it was uploaded to this 'AI Art' section.

We have some good artists here and this may be a way of preventing 'fake' images pervading our serious posts by literally giving those who want to play around with AI crypto images somewhere to show off their talent rather than try to pass it off as genuine. On the other hand, it may get to the stage where telling AI and real apart may be impossible. How do we manage the AI that is likely to eventually reach these very pages?

Thoughts? Other ideas?

Cheers,
Shazz

[Shazzoir]

Lordy, and I literally just found this to show what I mean...

AI.png

AI.png

[pencil]

First post here, but I think image metadata should keep us safe from fakes.
I don't think you can doctor metadata.
I hope everyone is doing well.
I got into the subject of hairymen about 2 years ago and have been hooked ever since.
Peace.

[Simon Park]

Shazz, you make an excellent point.

I'm deeply suspicious of most current images and video anyhow - and the entire 'deep fakes' thing is a good example of how easy it is to create convincingly lifelike images using modern technology. If someone has the correct computer knowledge, they can create some extremely believable illusions. You don't have to be a professional, either - the software that allows people to do this stuff is improving all the time.

Conversely, I'm sure that if someone really did come up with genuine video of something remarkable, the sceptics would waste no time in saying it was CGI.

It cuts both ways, I think.

They used to say that things were 'all done with mirrors'...now it's all done with computers.

[Shazzoir]

@pencil and Simon,
Thank you both for your thoughts
Good point about metadata; I hadn't thought of that aspect.
Simon, you're right about the constant improvement of machine learning; I think it will end up being a case of 'the servant has become master' eventually, not to sound like a conspiracy nut. Not sure if as a race, many of us are becoming lazier or if we're devolving ourselves, but when people around you start posting AI fakes or AI faked stories and claiming vehemently that they are the truth.... well. Uphill battle. Currently, AI seems to have trouble rendering hands and feet, as can be seen in the multi-digited examples I showed, but in time, it will figure it out because that's what it does; learn by collecting data in the form of images and text.

I wouldn't have a gripe with it if it didn't trick people into thinking it's real, or kill artistic expression by the hands and skills of a real human being who may have spent their whole life perfecting their art, only to have it bastardised by AI as part of how it learns to imitate.

AIso, I wouldn't mind as much if AI art was forced to label and tag itself as AI-generated, instead of presented as the truth, when it is not. So many do not have an enquiring mind, and believe everything they see on social media, which is problematic. To my mind, AI is the biggest plagiarism con. Sorry, I'll go outside to yell at a cloud now.
Shazz

[Simon Park]

I don't think you're in the "yelling at a cloud" stage, Shazz. Far from it.

There are uses for AI, but I agree with your point that it should be labelled as such. Pencil's point about the metadata is also a valid one but, sadly, most people don't even get that far - they take an image at face value and think no further. The scammers of the world succeed largely because of this fact.

I'll put a link below to a video that I think is a positive use of AI - but opinions are going to vary.

This guy's channel is extremely interesting and most people here are most likely already aware of him...but here's AI being used...well, intelligently.

https://www.youtube.com/watch?v=dMTloAMuq5E

[Simon Park]

Another video from M.K. Davis...


https://www.youtube.com/watch?v=kBawDTT34uU&t=116s

[Shazzoir]

And so it begins...

This has been all over the Thylacine pages in the last 24 hours, and seems to also have been deleted from the original Reddit post too.



The ego-stroking these people crave so desperately means AI will be used more and more. Fakery has a new level of existence, but I knew it was only a matter of time. We are no experts, but it could be AI, or more likely, AI combined with an image of one of the fibreglass Thyla models people have been using to fake sightings in the past.

Cheers,
Shazz

[Shazzoir]

Image with contrast upped and darkened a tad;

[Shazzoir]

Another case in point...

https://www.reddit.com/r/bigfoot/commen ... line_am_i/

AI makes it so you literally 'can't believe your eyes'...]

Shazz

[falke62]

Hi Shazzoir

If this is artificial intelligence, it must be very stupid. I'm not Australian, but Central European, but I recognized at first glance that neither the face nor the stripes resemble those of a Thylacine. So we don't need to be afraid of such stupid intelligence.
Have you heard this joke? How do you distinguish real Bigfoot photos from fake ones? Real ones are always blurry.

[Grantly]

With the rise of AI I've given up looking at any bigfoot/cryptid pictures anymore. Pareidolia is bad enough, now AI? I know Yoweis exist so I'm content with that.

[Shazzoir]

We are coming into an age where you will soon not be able to 'believe your own eyes' if it wasn't something seen in your real life.
This bloke even has a whole webpage about his AI Bigfoot images.
https://satanfudge.com/sasquatch/bigfoot-faces/

The age of disinformation is here.

Shazz

[Phunkninja]

I just thought i'd throw my two cents in here.

AI is at a point now where with just a simple text input, you can have a 4K video of a tabbycat walking through an alley.

With some training (which does not take long) you could re-create the patterson-gimlin footage with ease, if not with a trained model, have your own in your very own back yard.

AI will set any "valid" videos/photos back if not discredit them entirely. I've been learning about deepfakes, etc - its not that hard to do as long as you have a decent enough PC - so be super careful of any new yowie/bigfoot photos/videos you see.

[Shazzoir]

Absolutely. The Bigfoot groups online have been saying the same for the last four or five months. With this new Sora program, the gloves are off.

Of course, the company that owns it, 'Generative AI' is saying there will be checks and balances, but we all know how those tend to only apply in certain circumstances.
https://www.newscientist.com/article/24 ... -concerns/

By the time the legalities of using it are sorted, it'll be too late, the damage will already be done.
https://www.reading.ac.uk/news/2024/Exp ... s---expert