this post was submitted on 25 Jan 2024
67 points (79.6% liked)
Technology
59421 readers
3034 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
People have been creating and posting realistic looking fake celebrity nudes for quite literally decades now, but now they're using AI and its suddenly a problem?
I'm not sure if you noticed, but people who write for a living have suddenly started writing quite a lot about how technology that can write and generate media are bad.
Which is so silly, because AI writing still needs a human editor. I write for a living and there tons of work that involves using AI as a tool to increase productivity rather than to replace writers completely... like photoshop didnt put photographers out of business it just changed the work flow.
I work in a clinical setting where some Doctors are trying an AI program for generating their clinical notes out of the casual conversation between them and the patient. It's way off its mark for what we demand in quality. It requires significant editing from the healthcare provider, and if the note is very robust it quickly becomes more of a chore than modern voice transcription. Our review is not great so far.
That's a terrible way to be using a LLM for generating clinical notes.
Sounds more like trying to use a screwdriver to hammer in screws than an issue with the screwdriver itself.
This is what I think about AI being forced into many things these days. Feels more like an attempt to justify subscription plans than anything actually productive.
In part this is because the SotA model is by far GPT-4, but OpenAI has pigeon holed it into 'chatbot.'
The earliest versions of it pre-release when it was being incorporated into Bing were amazing. Probably the most impressive thing I've seen in tech.
But it was too human-like and freaking users out, so rather than wait for the market to adjust they did extensive fine tuning to make the large language model trained to predict human ouput be less likely to produce human-like output.
The problem is that they don't have a scalpel for this sort of thing and ended up with a model that's very good as a chatbot within a certain scope, but significantly impaired at some of the outside the box mechanics visible early on.
And because it's the SotA, everyone is now using it to fine tune their own models.
So the entire industry is being set back in practical applications outside of "kind of boring chatbot."
Right. It seemed like a reach when I first heard of it, but that's how it's advertised and the Hospital was sold on at least trying it out.
It's because a person can crank out a deep fake in 3 hours, and a crappy one in one. It never cropped up because... well lets be real it was a couple of weirdos that were doing it, unless it bubbles up from the dark corners of the internet you risk the Streisand effect by bringing attention to it.
AI can crank out 40 in a minute. 7200 in three hours. That's an entirely different beast. The sheer mass and volume ramps up the odds of any image bubbling up from the dark corners of the web falling into the limelight and now this problem that wasn't big enough to merit thought is rearing up it's ugly head right in front of us.
You can generate unique pictures of Taylor Swift faster than even Taylor swift can generate pictures of Taylor Swift. Within one hour of Taylor swift being seen with a man (and you have enough images of the man) you can create a dozen images of her on a date with that man and attempt to sell them to paparazzi.
The problem is volume. Just like how email made everyone connected and allowed the Nigerian Prince scandal to occur.
It's also not just limited to Taylor swift.
People don't care now because people don't thin famous people should have any right to dignity, but their minds are gonna change really quickly when it's their sister, mother or daughter.
I don't think many have gone viral on social media before or took less than 5 minutes to create. My uneducated guess is that previously this stuff would be in some niche forum in the recesses of the internet
Not really. Back at the time there were public usenet groups specifically dedicated to the [hot actress of the day] fake porn.
I can see why people are upset, I can agree that distribution of these images can be an issue, but this has the same energy of "I am mad that a certain picture of me is on the internet, I demand that they take it down." Sorry, that's not going to happen any time soon.
You could be honest and acknowledge that there is a massive difference in time investment and skill required between the old way of creating fake porn of unconsenting people and the new way.
It's now massively accessible and realistic. Yes, it's a problem.
Google.com/images
Damn look at that, been accessible for decades
The tools are accessible. I wish this place wasn't full of weirdo ai tech bros sometimes.
Because downloading gimp is really hard.
I'm sure you do not need me to explain to you the cavernous difference between gimp and deepfake. I trust you are capable of that.