But how good is it really . I think fake. That but how good is it really . I think fake. That is but how good is it really . I think fake. That is actually| think fake. That is actually real. Think fake. That is actually real and think fake. That is actually real. And joe think fake. That is actually real. And joe investors think fake. That is actually| real. And joe investors the real. And joe investors the darker side image creation. Darker side image creation. Investigates. This month fans all over the world were glued to coverage of the wimbledon tennis tournament. It to coverage of the wimbledon tennis tournament. Tennis tournament. It was televised tennis tournament. It was televised for tennis tournament. It was televised for the tennis tournament. It was televised for the first tennis tournament. It was| televised for the first time tennis tournament. It was i televised for the first time in 1937 when fred perrys run of wins was ended by the american don budge. Wins was ended by the american don budge don budge. Since then the bbc and broadcasters don budge. Since then the bbc and broadcasters around don budge. Since then the bbc and broadcasters around the i and broadcasters around the world have provided Expert Commentary on all of the action. Well, actually, not quite all of it. Quite all of it. Covering of the outlying quite all of it. Covering of the outlying cords quite all of it. Covering of the outlying cords has i quite all of it. Covering of i the outlying cords has remained somewhat silent, but they could be about to change thanks to Artificial Intelligence as Shiona Mccallum has been finding out. Shiona mccallum has been finding out. What was it like as a player finding out. What was it like as a Player Walking as a Player Walking out on centre court . It as a Player Walking out on centre court . As a Player Walking out on centre court . It is ust the most Fantastic I Centre Court . It isjust the most fantastic feeling. In | most fantastic feeling. In front of this crowd, touring year 1 and front of this crowd, Touring Year1 and that thrill and bit of fear as well, but all mixed in and there is nothing like it, itsjust in and there is nothing like it, its just wonderful. In and there is nothing like it, itsjust wonderful. She it, its ust wonderful. She was it, itsjust wonderful. She was britains it, itsjust wonderful. She was britains number i it, itsjust wonderful. She was britains number one | it, itsjust wonderful. She was britains number one in the 19805. Was britains number one in the 1980s. Like other players and fans who knows what a magical place wimbledon is. Nothing beats the atmosphere here on centre court in the sunshine in wimbledon. It is epic. But not everyone can make it to south west london. For a lot of us that means watching on television, mobile or computer. Not all of which will have commentary. The All England Club has been working with its it partner ibm to fix this for the wimbledon website and app. Neglects masses of data from all around the courts for its ai platform called watson x. We take the ai platform called watson x. Take the information fitted into the model which then takes text, turns it into numbers, tasty numbers and compresses it, expanded, then outputs tests again, which can then be spoken as an Ai Commentary. fiend spoken as an Ai Commentary. And this is what spoken as an Ai Commentary. And this is what it spoken as an Ai Commentary. And this is what it sounds like. Iga swiatek, ranked first. Both male and female sounding commentators are featuring the app. Commentators are featuring the a. ~. ,. , app. Murray, from the united kingdom. App. Murray, from the united kingdom, will app. Murray, from the united kingdom, will play app. Murray, from the united kingdom, will play tsitsipas. | kingdom, will play tsitsipas. It was the boys are Computer Generated which is why they lack, how can we put it, they lack, how can we put it, the human touch. The human touch. What a fantastic the human touch. What a fantastic place the human touch. What a fantastic place to the human touch. What a fantastic place to do i the human touch. What a fantastic place to do an l fantastic place to do an interview. It is amazing, right on centre court, sons out. Ia on centre court, sons out. 10 has on centre court, sons out. Jo has been commentating since she gave up playing. You has been commentating since she gave lip playing gave up playing. You need to be short and sharp gave up playing. You need to be short and sharp sometimes i gave up playing. You need to be short and sharp sometimes with i short and sharp sometimes with what you are saying. And you get a feel of what is happening. Sometimes you dont need to say anything because the crowd have gone mad and you want to feel that and see the players. So let it go. Other times, maybe when it is flagging a bit, it is time to fill in a bit from behind the scenes. So i think you very much have two have that feel of when to talk and when not. So is thisjust the when to talk and when not. So is this just the start of Computer Generated commentary replacing people likejo durie . This is on matches that dont have commentary at the moment. We would love to make all matches available in a more accessible way and you can see how the intention here is to complement and supplement those commentators that we know and love rather than replace them. For now the Ai Commentary is only being used after the event, for highlights, but ibm has plans to generate commentary in real time. It has plans to generate commentary in realtime. It is a net approach. Commentary in realtime. It is a net approach, diwali, commentary in real time. It 3 a net approach, diwali, the types of shots. A net approach, diwali, the types of shots. A net approach, diwali, the types of shots. But how close to the types of shots. But how close to the goal types of shots. But how close to the goal of types of shots. But how close to the goal of a . Types of shots. But how close to the goal of a . We types of shots. But how close to the goal of a . We are i types of shots. But how close to the goal of a . We are very i to the goal of a . We are very close. These to the goal of a . We are very close. These models to the goal of a . We are very close. These models are i to the goal of a . We are veryj close. These models are very big. I will tell you a very interesting statistic. By the time, you know, attempts point is over, i can create commentary in 1 2 seconds. It is really fast. At the beginning of the project, that was the grand challenge, to be as fast as possible, and we were at the order of 20 seconds. Were at the order of 20 seconds were at the order of 20 seconds. ~ ,. ,. ,. , seconds. Will be able to do it in less than seconds. Will be able to do it in less than a seconds. Will be able to do it in less than a second, seconds. Will be able to do it in less than a second, is i seconds. Will be able to do it in less than a second, is that| in less than a second, is that the aim of this . The aim of this . Gas, so i think it the aim of this . Gas, so i think it is the aim of this . Gas, so i think it is achievable, i the aim of this . Gas, so i think it is achievable, for| the aim of this . Gas, so i i think it is achievable, for us to get there. We could hose skill at some of our algorithms. Yes, so. We could make a Data Collection faster as well. 50 could make a Data Collection faster as well. Faster as well. So when my eo le faster as well. So when my people be faster as well. So when my people be watching faster as well. So when my people be watching live i faster as well. So when my | people be watching live and getting real time Ai Commentary . It getting realtime Ai Commentary . Getting realtime Ai Commentary . Getting realtime ai commenta . , commentary . It will be sooner than any of commentary . It will be sooner than any of us commentary . It will be sooner than any of us could commentary . It will be sooner than any of us could really i than any of us could really imagine. But it only roll around to next years tournament i think you will really enjoy what you see. Really enoy what you see. When ou see really enjoy what you see. When you see this really enjoy what you see. When you see this technology really enjoy what you see. When you see this Technology Moving | you see this Technology Moving in the future and do you think it will be on real time matches, for example, next year . I matches, for example, next ear . � matches, for example, next ear . , year . I cant give you any commitments year . I cant give you any commitments but year . I cant give you any commitments but i i year . I cant give you any commitments but i can. Year . I cant give you any i commitments but i can say what we are going to be doing is innovating and inventing new experiences which ab we havent even imagined yet. {lilia experiences which ab we havent even imagined yet. Experiences which ab we havent even imagined yet. Ok, but what do the professionals even imagined yet. Ok, but what do the professionals think . I do the professionals think . Well, now were going somewhere where it is a bit scary, i think. For me, for my mind, cant quite understand how an ai would get the feel of the situation. There might be certain things that happen in a match that you can say they missed this or that, but what about the reaction of everything because . What about the reaction of the player . What has just happened because they to get my head around this, i am just not sure. And will i be out of a job soon . I dont know. Will i be out of a ob soon . I dont know. Will i be out of a ob soon . I dontknow. � ~ , � , dont know. Ibm says jo duries ob dont know. Ibm says jo duries 0b is dont know. Ibm says jo duries job is safe dont know. Ibm says jo duries job is safe for dont know. Ibm says jo duries job is safe for now. Dont know. Ibm says jo duries job is safe for now. But dont know. Ibm says jo duries job is safe for now. But five, i job is safe for now. But five, ten years from now, well, but as a whole different ballgame. That was shiona and i know we are a week on now but i am still reeling from both of those wimbledon singles finals. To be honest, i think, right now, i would to be honest, i think, right now, iwould rather to be honest, i think, right now, i would rather hear a given commentator attached to those amazing shots. That was amazing tennis. Those amazing shots. That was amazing tennis. Yes, and aside from some amazing tennis. Yes, and aside from some of amazing tennis. Yes, and aside from some of the amazing tennis. Yes, and aside from some of the amazing i from some of the amazing opportunity that al brings us, it can also have a dockside. You may find this next piece hard to watch, but it does show the reality of what is happening in the world of ai image generation. Jae happening in the world of ai image generation. Happening in the world of ai image generation. Joe tidy has been investigating image generation. Joe tidy has been investigating the image generation. Joe tidy has been investigating the story i been investigating the story and is part of it he has been shown redacted versions of some of these images. You wont see them here, butjust to repeat you might find the next film upsetting. Joe tidy. You might find the next film upsetting. Joe tidy. These da s upsetting. Joe tidy. These days anyone upsetting. Joe tidy. These days anyone with upsetting. Joe tidy. These days anyone with a upsetting. Joe tidy. These days anyone with a laptop i upsetting. Joe tidy. These | days anyone with a laptop can be an artist. With days anyone with a laptop can be an artist. With al and an idea, these sky is the limit. And online galleries and forums are exploding with creations. By are exploding with creations. By typing and words you can create any image that you like, so, for example, a Cat Holding A Banana in the style of van gough, lets try that. The only real limit is your imagination, but there is your imagination, but there is your imagination, but there is your imagination, but there are safeguards in place. Van gogh. Many of these tools have bad words or phrases to stop people creating illegal or offensive imagery, but we now know that hasnt really worked. There are only three organisations in the whole world who are licensed to actively search for child sexual abuse material. And this is one of them, the Internet Watch Foundation here in cambridge. Its highly secured facility, the iwf fines, reversed, and log some of the most abhorrent photos and videos are measurable. They see a flood of images made with Artificial Intelligence. And we are the first people out of the charity to c redacted versions of some of the images being shared by predators. The first imate i shared by predators. The first image i want shared by predators. The first image i want to shared by predators. The first image i want to show shared by predators. The first image i want to show you i shared by predators. The first image i want to show you is l image i want to show you is more of a cartoon style of image of a girl on the beach. I would say she is about 3 6 years old. It would say she is about 36 years old years old. It is really realistic. Years old. It is really realistic. Is years old. It is really realistic. Is quite i years old. It is really i realistic. Is quite realistic. If i saw that i would think it was a photoshop damage. This location is was a photoshop damage. This location is a was a photoshop damage. Try 3 location is a bedroom and that is quite classic of real images that we see of children posed in the bedroom. D0 that we see of children posed in the bedroom. That we see of children posed in the bedroom. Do you see that e of in the bedroom. Do you see that type of image in the bedroom. Do you see that type of image. Yes. In the bedroom. Do you see that type of image. Yes. Children i type of image. Yes. Children that young . Type of image. Yes. Children that young . Yes. Type of image. Yes. Children that young . Yes. It type of image. Yes. Children that young . Yes. It is type of image. Yes. Children that young . Yes. It is a i that young . Yes. It is a horrible that young . Yes. It is a horrible image. I that young . Yes. It is a horrible image. Yes. | that young . Yes. It is a horrible image. Yes. Children that young . Horrible image. Yes. Children that young . Yes. Horrible image. Yes. Children that young . Yes. It horrible image. Yes. Children that young . Yes. It is horrible image. Yes. Children that young . Yes. It is a that young . Yes. It is a horrible that young . Yes. It is a horrible image, that young . Yes. It is a horrible image, the i that young . Yes. It is a | horrible image, the idea that young . Yes. It is a i horrible image, the idea of what she is doing here. I. Hard to see, actually. It what she is doing here. I. Hard to see, actually. Hard to see, actually. It is difficult. Hard to see, actually. It is difficult, absolutely. I hard to see, actually. It is| difficult, absolutely. Sorry. Sorry about difficult, absolutely. Sorry. Sorry about that. Difficult, absolutely. Sorry. Sorry about that. Ive i difficult, absolutely. Sorry. | sorry about that. Ive never experienced anything like that, and its a shock. And its a shock. Absolutely. And there and its a shock. Absolutely. And there is and its a shock. Absolutely. And there is no and its a shock. Absolutely. And there is no need and its a shock. Absolutely. And there is no need to i and its a shock. Absolutely. And there is no need to be i and there is no need to be sorry because these images are shocking. Sorry because these images are shockina. , sorry because these images are shockina. ,. ,. , shocking. Last month for the first time shocking. Last month for the first time the shocking. Last month for the first time the m shocking. Last month for the first time the iw have shocking. Last month for the | first time the iw have studied actively logging reports of ai images. Analysts discovered galleries are multiple websites, some containing category a material, the most graphic possible. Iwf. Stable diffusion is the most popular open source image generator, that has been repurposed and repackaged by countless businesses, giving microsoft another is a run for their money in popularity and power. But Stable Diffusion doesnt have a Silicon Valley start up origin story. It started here in a leafy Downtown Corner of munich. And the way it was launched and created was completely unique to any of the other ai image generators. To any of the other ai image generators generators. We are all now really looking generators. We are all now really looking over generators. We are all now really looking over the i really looking over the shoulder of a Development Process which is super rapid. Professor bjorn was the lead scientist on the Stable Diffusion project. He and his team did their best to remove pornographic content from the pornographic content from the pornographic images of the ai pornographic images of the al was trained on, but he admits it wasnt perfect. They also coded in a list of hundreds of bad words and phrases, but of course people quickly found a way to delete them. He defends their decision, though, to unleash his model onto the internet is open source. The Dancers Internet is open source. The dangers that internet is open source. The dangers that you internet is open source. Tue dangers that you betrayed internet is open source. Tte dangers that you betrayed with open source, yes, isee overall, like, there is potential and generative ai, it is a powerful technology and powerful can be misused, but closed source has not proven to be the way that would actually do this mitigation for us, because it has been either leaked or more importantlyjust been re implemented. But leaked or more importantly ust been re implemented. But you must accept been re implemented. But you must accept that been re implemented. But you must accept