AI

Been saying this for the last 5 years about AI. Be afraid, be very afraid!!

The vast majority of people have idea about what it could do and will be used.

Follow the money. Find out who owns Google, Facebook, Twitter, and all the other social media sites viewed. Then find who sells AI apps, designs the way they work, and defines the databases on which they are trained.

These online media are by far the biggest influencers on the internet, and thence the biggest influencers on the vast numbers of people who, knowingly or not, use the internet for international news and information on political trends as well as for which air fryer to buy - which shapes how they see the world and how they vote.

But you know which influencer is pressing you to buy what items - and you know that they choose what will bring them the highest income. AI is anonymous and bears the face of neutrality.

With google searches you get many pages of repeater content often with one bias or promotional aim, Asking AI a specific question gets a much quicker result, of course you have to ask the correct question.

An AI model could be built with a bias but that’s no different than individual source bias that we get in the media today from so called “influencers”

For Facebook I also mean Meta.

foxie48I wasn't commenting on the quality of you photographs. I am sure they were of top quality and not AI-created - and the NHS is not likely to be faking the images by which it gets a diagnosis. The algorhythms learn more as they get feedback on their work, so the more diagnoses they do, the more accurate they become.

My point was that AI apps in general are trained by giving them a very large database of examples of the work they are to be doing, taken from what has been done and published in the past. They refer to those examples when asked to answer queries, diagnose diseases from photographs, design houses, cars, interplanetary spaceships, furniture, fashionable clothes, nutritious meals, complicated electronic circuits, etc, write articles on all kinds of subjects, and write content for the never-satisfied demand of the public for "input".

Facebook has stated that the material on which its AI bots are to be trained is the whole of whatever has been added to FB by members and posters! There was a stushie when FB posters realised that it was to be assumed that if they had mentioned anything on Facebook, they were happy for it to be used by the AI as part of what it was creating or searching or compiling in a database, without asking permission. That is, names and locations, photos including children, connections, opinions, anything. Old photos are an eyecatching illustration to a FB post, so AI uses a lot of them. Unfortunately their research for the captions they write explaining what the photos depict is flawed or non-existent. It suffers from them not being human and not understanding what they see, it is just patterns of light and shade which they have been "taught" mean certain things. For instance they describe cobbled streets very frequently, when the photo shows perfectly smooth ground - because somewhere they learnt that Victorian streets were cobbled.

The material created and posted by AI was/is added to the total Fb content - increasing the size of the "training" database. So the AI was/is being trained on the content of its own products. Since a lot of the original content was dodgy in accuracy and bias level to start with, repeating it doubles up the inaccuracy and bias and "creates" nothing.

My guess is that most elderly people will be intransigent about applying themselves to learn how to safely use artificial intelligence machines.

I have used a couple of AI searches recently after being unable to get a decent answer on a search. Results I got were surprisingly comprehensive.

foxie48

*Elegran*f I think we may be talking at cross purposes. The photos were taken by a technician using a special high resolution camera at a clinic. These were then uploaded and analysed by an AI machine located somewhere else, which then generated a report for the NHS. I've no doubt that the photos of my lesions will be used to train the machine with the correct diagnosis. I've absolutely no idea why it failed to correctly identify my lesions but it is still in the early stages of being used in the NHS and it should save money and time whilst ensuring that patients receive timely treatment for the most urgent skin cancers.
Having had this experience I would urge anyone seeing something new on their skin to get it checked out. The lesion on my nose was not obvious but it started to bleed, initially I though it was just a scratch and being on blood thinners, I tend to bleed easily. I'm so glad I got it checked out fwiw, the nose is not easy to work on surgically as there is little flesh and no extra skin so it's turned out to be more problematic than anticipated, but at least the cancer has gone. I just need the wound to start healing!

You are completely right- people seem to underestimate how serious these bccs can be. I found the tiniest little mark on my mums neck.. 18 stitches she had.

I’ll just stick with old fashioned ways of doing things. I don’t trust this AI stuff.

Elegranf I think we may be talking at cross purposes. The photos were taken by a technician using a special high resolution camera at a clinic. These were then uploaded and analysed by an AI machine located somewhere else, which then generated a report for the NHS. I've no doubt that the photos of my lesions will be used to train the machine with the correct diagnosis. I've absolutely no idea why it failed to correctly identify my lesions but it is still in the early stages of being used in the NHS and it should save money and time whilst ensuring that patients receive timely treatment for the most urgent skin cancers.
Having had this experience I would urge anyone seeing something new on their skin to get it checked out. The lesion on my nose was not obvious but it started to bleed, initially I though it was just a scratch and being on blood thinners, I tend to bleed easily. I'm so glad I got it checked out fwiw, the nose is not easy to work on surgically as there is little flesh and no extra skin so it's turned out to be more problematic than anticipated, but at least the cancer has gone. I just need the wound to start healing!

butterandjam

Most of us are too old to get pregnant. Though my door is always open to George Clooney

😂

"AI training data is a collection of information used to teach AI models to make accurate predictions or decisions. This data can include labeled images, text documents, audio recordings, and sensor data. The quality and quantity of training data are crucial for the accuracy and efficiency of AI models" AI generated!

Another area in which AI may be dangerous, is the world of online dating.
In an area where people are already not that honest, another level of dishonesty has been created.

Sorry for what you went through foxie48.
I am not at all suprised it was wrong in the area of dermatology.
I am under a dermatology Consultant.
That area of medicine in the NHS is serioulsly flawed. In my opinion. In lots of different ways.
Or perhaps it is nor as advanced as some areas of medicine.

That isn't data, though, foxie That is the machine's analysis of what it saw. The photographs are the data. - the raw material on which the analysis is done.

Facebook is full of photographs - genuine ones taken in the many years since photography was first invented. Photographers have always been selective in what they photographed, and what angle they stood at at take the picture. In the wet darkroom they emphasised some parts with more light and put others into shadow, and they could manipulate images with double exposures and other techniques. however, the manipulation took time and skill, as did the reproduction of pictures. Then with digital cameras there were photographs edited in Photoshop and other apps, these were faster to do and came with readymade effects and techniques.

AI "photography" is as far ahead of Photoshop as Photoshop is of a Box Brownie and a roll of Kodak film. The AI apps can take a few snaps of someone and turn them into a video of them ten years older or younger, in different clothes, chatting to someone else copied from an existing image (or invented) and put words and opinions into the conversation that neither of them ever said in real life. Also an AI caption can be completely mistaken about what is being pictured in a scene, .

The results are twofold - that the cloned "deepfakes" are believed by many to be real, which can smear the people in them and cost them their jobs, reputations and lives, and that people then don't believe real photographic records - or handwritten contemporary accounts - of things that happened in the past, or videos of atrocities being committed.

AI is only as good as the data though. I recently used the AI dermatology service which is used by the NHS in my area. I have high resolution photos taken of two areas of skin that the GP was concerned about. Within two weeks I had been assessed and had an appointment with a human dermatology consultant who assessed the two areas in literally two minutes. The AI results were completely wrong. What was assessed as a fairly benign pre cancerous area on the bridge of my nose was in fact basal cell carcinoma, the area on my arm was not squamous cell carcinoma needing urgent treatment but it did get me in front of a dermatologist quite quickly. I've had the bcc removed privately as the NHS wait was at least 9 months, in which time there was a risk I would have needed much more invasive surgery. AI will be a huge help in medical diagnosis but it certainly isn't perfect (FWIW my GP diagnosed correctly).

Sorry that is so long. It started as agreement with Oreo and somehow turned into almost an article. I'm not even dressed yet.

I marvel at the AI videos on social media that show situations that just aren’t possible. For example, a dog, cat, rabbit and bird peacefully sitting in front of a fireplace together. It’s obvious to me that it’s AI but the amount of people who are fooled! Comments like “what a beautiful video, showing the love between animals.”

Oreo

AI is here to stay, I think we can all comprehend the dangers and the positives of it.What it does need is regulation.

Artificial Intelligence (I wish its initials were not the same as those of artificial insemination) has immense power for good and ill, and power without responsibility is a dangerous weapon.

On the side of the angels, AI can absorb factual information and categorise the data and search through it for statistics and conclusions in nano-seconds which would take a human being a lifetime of work. It can carry out a surgical operation more accurately than a top surgeon, and do it from thousands of miles away. It can operate a vast warehouse using automatic vehicles that locate items, retrieve them, pack and despatch them. It can manufacture items from basic materials to finished product.

All of these require the AI system to have minute information in the form of a database of the measurements, spatial co-ordinates, chemical nature, functions, and so on of what it is physically working with, or of the theoretical base and the history of the philosophies and debates which it is analysing for a decision on a political change or the previous history of an individual for whom it is planning to pursue a criminal charge.

However, it makes draconian decisions, and can't be questioned on them, reminding me of the definition of a bureaucratic committee - it has no heart to appeal to and no backside to kick.

These databases have been created, assembled, added to and stored by human hands, with all the potential for mistakes, omissions, unremoved out-of-date facts, and bias or downright malice on the part of the collator. Then, once they have digested by the AI system, they become part of the system's "reality". - on which life-or-death decisions are made.

A national police force is also a power for good or evil, but in most countries there are restraints on how that power is exercised, so that it can't be used as an instrument of political oppression, and internationally, there are restraints on how one country can use its army to control a separate country.

However, AI is invisible. It doesn't roll up at 3 am in an unmarked car to arrest a sleeping family on a charge of some vague misdemeanor and shut them up, never to be seen again. It doesn't fly above the clouds in a formation of military jets and bomb the bejasus out of a foreign country. It doesn't rob banks and carry sacks of money and jewels to a robber baron's hideout. Yet those who control the data which fuels the country's AI systems and the programming by which the AI engines run can achieve the same results, by influencing the assumptions on which votes are cast and decisions made.

In the service of the propaganda department, AI (or rather, the users of it) can manipulate the masses of images, videos texts, articles, that have been collected and use them to win the hearts and minds of the population of the country it wishes to control. There are more ways than one to control the views, allegiances, votes and particularly the spending power of a population. In these days of almost instant global trade, it isn't necessary to invade a country physically to plunder its wealth and resources. You can do it while they sit watching TV or surfing the web.

Whiff

Caleo if you just put AI as a title you will get confused posts .

So what do you mean will AI be used to use as a weapon against other countries like nuclear bombs or to de stabilise a countries economy ?

Absolutely this AI covers so many different things. I use AI for work, to do grant applications, to write letters and so much more, it is brilliant. AI can be a hundred things, you need to make yourself clear what you are talking about OP.

AI is seriously scary in the wrong hands. And it WILL get into the wrong hands. I’ve long wondered how the world will end. Nuclear war? Global warming? No, it’ll be AI. That’s for sure.

The Times "Story" podcast was excellent yesterday, "One day in the life of an AI chatbot". I found it quite scary but really worth listening to. Do we really understand what we are creating? tbh I'm not sure we do.

So, hopefully we are all wise enough to verify anything that Insta, FB or TikTok has to say already.

AI in Google is just a summary of all that, but it also includes govt website links and Wiki info, so can be a shortcut, but hopefully we all verify the same way.

It is really good for redesign - you can change the colours on images, and for image searching.

However, like ALL tools, the danger is that folks rely on them to make decisions instead of simply using them as indicators - but we have been doing that since computers became prevalent.

.

You don't get irony then?
Are you American!