Just read this item and thought of the implications now for everybody…
We all get mail, spam, phishing scams, etc. With the usual ideas, you have won, you have a package to be delivered, help needed, fred has died and left …
These scams are widespread and I probably get a good number each day headed direct to the bin.
But with the growth of artifical intelligence, voice recognition, speech and images these types of scams could get worse. Imagine a image of a loved one on the screen interacting with you asking for help and matching lips to voice to the plea and being able to answer your questions as it has been listening in through your mobile device or smart tv…
We alread have pop-up blockers for ads, spam catching systems. Now do we need AI catchers ?
Brave new world!
I saw a guy demonstrating AI capabilites earlier this year.
He just took a short video of himself talking for a minute.
Then AI generated a video of him talking about anything, basically written in a text.
And this was a live demo!
After “speaking” about the subject in english, just a mouseclick, machine translated text to french, then to german, and couple seconds later he spoke about the same subject in french and in german -on his own voice!!!- however in real life he doesn’t know a single word of those languages…
The audience applauded, there was a big success. However I find this rather scary…
I imagine that is difficult for lip synchronised movement especially as english french and german move words around to form a phrase.
If i write something and then use google to translate into french, then my wife corrects it to better format, something of one line can easy become 2 or 3 lines… i imagine german is the same
Mouth movements were in sync with the sound, just like the head of the guy, as speaking the subtle turnings of his head, even eyebrow moved realistic on his face…
It was surprinsingly shocking realistic fake video done in seconds.
The only drawback that was apparent that time this did not work for hungarian language.
I’m not sure it is still the case.
I don’t speak french at all, so I can’t tell the how good the translation was, but it sounded very nautral, I understand still a little german, and that felt OK.
But anyway, if the text needs some corrections, it can be done before generating the video.
It’s not that the translation was good or not, but a written text in different languages can be spoken by any impersonated human (assuming having enough sample of his/her voice, and image).
Imagine Trump encouraging people to vote for Kamala. Or the other way around, doesn’t really matter. It is doable with this technology, and it would be really very hard to tell it’s fake.
I’m scared of this tech.
Interesting that this topic appeared today. I was reading on one of my game forums yesterday and one of the Canadian members received a scam phone call using this technology. I did not even know we were there yet.
The gist of it was:
Grandson calling saying “This is your grandson. I had a wreck and now I have been taken to jail and need a lawyer and bail bondsman to get out. Please help me!”
She was an astute tech person so she replied, “What is your name?” But the AI voice kept replying, “It’s your grandson! Help me!”
Bottom line, she knew that even though it sounded just like her grandson, it was not like him to call in this way and that not providing the name, was a red flag. Obviously, a video might be more convincing.
To me, the whole AI thing is scary. Given the voice scams that will now bombard us along with the email, text scams. SMH.
But as @nevj said, AI can be very useful in debugging computer issues. And I have seen that evidenced in some of @Rosika threads as she would give us ChatGPT answers.
I recently used Duck Duck Go AI in searching how to do something in Linux and got a very concise, correct answer.
This is a worry, imagine older people or those not aware of the scamers doing this.
Earlier this year, i had a call from a client in panic about a virus issue could i go immediately, no problem, but when i arrived with my wife who chats with the client whilst i work on the problem.
A couple about the same age as us, intelligent and aware of things. Had a call from tge bank about an unusual transaction and needed access to the computer in the home to check something, which they agreed too, Friday evening strange, saturday morning when trying to use the card everything refused. Hence the call to the bank to discover they had been scammed. Bank stopped the cards immediately and sent them to the police to report the crime. The police recommend scanning the computer and to call me. The scammer had been on the machine, stolen all card details and withdrawn money. Virus everywhere key loggers, and a real mess.
To solve it I took the drive out, replaced it with a new linux system and just copied files back on. Plus went through every account they had changing passwords. Email, amazon, etc.
The old hard disk they handed into the police for the crime team to search for the fraudulent actions. Thats why I changed disks, system also to reduce the risk as I think its harder in linux to do these types of crime.
Sadly they were not alone, the local police along with a older peoples team set up a conference to warn others in the area, with over 150 attending the conference hope the message gets through.
I hope my texts won’t be used to train AI or to replicate me if they get published.
On another note: I have a very dim view of writers using AI to write their texts, or even using AI for editing them. Writing is an art, and art is done by humans, not by machines.
Mills and boon (hallmark) have been doing it for years.
Boilerplate 100 paragraphs, change the names and location… here comes another formula hit, 350 pages, 20 chapters, girl meets boy, falls in love, shock happens, they seperate, then get back together. Dont need AI for that formula.
The TV films are the same.
I promise I have never read one just wait to be corrected
It seems like universities don’t like ChatGPT at all … anything from it … so much so that they have shit like TurnitIn which will detect stuff coming out of ChatGPT … I think they’d rather I use wikipedia as a source… yet, nowadays, whenever you do a search, the first thing you get is a chatgpt generated answer… What gives?
Glad i have stepped away from teaching now as it must be really hard to evaluate research as being original and not just copy paste from so many internet sites. Or artificial intelegence.
perhaps that is the future of teaching replace professors with chat gpt ?
University people used to say
" You need to learn how to find out things"
which, in my day meant, how to use the library.
Everything in the library is not on the internet… Google tried to do that and was prevented
So are today’s students, ie you, able to use library search techniques to find old stuff not digitised?
How much of your thesis review comes from non-internet sources?
I cant make up my mind whether universities are just being conservative, or whether it is still really important to learn to use a library. I am too long out of it.
Using stuff written by someone else, without acknowledgement, is regarded as plagiarism. If that someone else is ChatGPT, it is still plagiarism.
It would be theoretically possible to ask ChatGPT to write every possible sentence on every topic it could think of…
would take a bit of storage.
Then there would be no sentence left for you to create as your own… they would all already exist.
So there would nothing left for academics to write/say.
Can you see why that threatens Universities?
It is worse than that. You could generate every possible combination of pixels in every video frame of every movie that could ever be made. Somewhere in there would be an exact copy of “Gone with the Wind” for example.
Can you see why that threatens movie-makers?
There has to somehow be room left for human creativity.
We are all makers. We like to build/create things.
When everything is built, what is left for humanity?
Is it capable of something new ?
Like the industrial revolution, or the changes that have taken place over the last 100 years in way of technology… think radio to TV, library’s to internet, processors, memory the list goes on.
Could AI invent a replacement for petrol, or an engine replacement as man has done.
I think no.
It might be able to find a new application for something that already exists.
But AI is not a ‘maker’. It does not create things. It could never have written “Alice in Wonderland”
Historical events are not invented by individuals… you might call them social inventions… they are like “emergent properties” of aggregates… eg the insulating behaviour of a woolen garment is not a property of a single wool fibre… it emerges when you put fibres together.
The industrial revolution happened because people did things together. I dont see how AI could organise people.
