Creative Chat - How Artificial Inteligence will reinvent who we are
Added 2019-05-23 16:55:05 +0000 UTCWe are at the brink of an artistic revolution. One that will call into question what art is, what creativity is and what makes us human. We already see the beginnings of this with AI being used in more and more Creative Software tools to create designs and artworks in seconds that would use a lifetime of training to reach that level of skill. Most of them still only simulate creativity based on styles invented by humans (like in style transfer), but we already see AIs that can create creative styles from scratch. Let's talk about how this will change Art, Photography and Photo Editing.
Photo used:
https://unsplash.com/photos/YKW0JjP7rlU
AI Used: https://deepdreamgenerator.com
Source Links:
https://artsandculture.google.com/project/ai-more-than-human
Nvidia AI https://www.youtube.com/watch?v=hW1_Sidq3m8&t=2s
Style Transfere to Video https://www.youtube.com/watch?v=Uxax5EKg0zA
Support me on Patreon:
https://www.patreon.com/sarikas
Follow me on:
https://www.facebook.com/sarikastutorials
https://www.instagram.com/sarikasat/
Join us on Reddit:
https://www.reddit.com/r/AffinityPhotoTuts/
Comments
Hi Kevin, those are some great points. You are right, we have dreamed about this for a very long time. I think the big surprise is that the AI we have now is so very different from anything we imagined in the past. It's actually a Black Box that is so complex in what it does that we still have no way of understanding how AI actually reaches it's result. At the same time, if we look at our own creative processes, a lot of them a based on math and rules. Music, Illustration, architecture, or software. Since math is the language of computers and they can use it much faster than we can, they already have a huge benefit. That leaves out what creativity actually is. So you are totally right that AI creates variations of our creative things instead of being genuinely creative. But at the same time, we ourselfs are merely creating variations of what has been done before as. Virtually no artist ever creates somethings genuinely new out of nowhere. So, there really is the question how big the difference is between the two and what it would take to melt that difference away. At the moment AI is very clumsy because it goes by random experiments and goals of what qualifies as good or bad set by us. But i can imagine that a more specialized AI can get a deep understanding of how and way for example different note combinations make for a better music, make it more happy, more sad, more artistic or more entertaining. At that point we have two major problems: 1) is AI now actually more creative than we are and 2) could a company build such an AI create millions of notations that are pleasing to us and copyright them all, so every future artist would have to play royalties to them. This is such a big can of worms we are about to open. But i love to look at the positive side of things and also want to see what's going to happen.
Olivio Sarikas
2019-06-06 10:39:36 +0000 UTCHi Olivio, In my mind the progress of AI has been a long time in the making. This has been in the human imagination for centuries. It is only now that humanity has the beginnings of the technology to start to make it happen. There is no doubt that AI can assist human endeavour but I doubt that it could do more than that. One theory of the industrial revolution that machines could do the work of people so that they could have more free time. AI will probably continue that theme for the next few decades. In terms of AI art creativity, it will be able to do that in spades BUT it will need a starting point - some inspiration, if you will. In my opinion, this is what will always be lacking in AI. It is not sentient as we or the animals are. As a human race we are unable to define was sentience actually is, so it is unlikely that we will be able to programme it into a section of code. There are some very good systems that have (or get close to) passing the Turing test, but they are not sentient. Of course, as Pete and yourself have been discussing, humanity may be creating it's own monster here. But that is not the fault of AI, rather the fault of humans not learning the lessons of history.
Kevin Phillips
2019-06-06 10:06:40 +0000 UTCHi Pete, i think it is still worth the risk ;) here are two good videos about this topic https://www.youtube.com/watch?v=48mf2QUtUmg and https://www.youtube.com/watch?v=QPwhEnAILa0 I think unless there is a really dramatic event that will force us to put rules and limitations into action, we will go for AI and there are already a ton of AI supported products out there. It's a really interesting topic and it can go both ways. At the same time it is very hard to judge, because it is a form of technology like we have never seen before. It certainly will dramatically change our society and bring dramatic improvements in science and art. Let's hope for the best outcome.
Olivio Sarikas
2019-05-30 14:27:52 +0000 UTCNo problem Olivio! Yup, this AI business is a rabbit hole :) I think there is a risk of us underestimating the capability of AI to be destructive of society. I agree its much more likely that extreme actions would be initiated by us humans (by design or otherwise); history is littered with such occasions and as our destructive power has increased, so have the consequences - more than once, a slight difference in approach or decision could have led to terrible events. My point is, that whilst AI might not have the sort of 'self-awareness' attributed to humans (I put it that way as our "existence" is a difficult to prove philosophically - many human traits, skills, etc. can just as easily now be attributed to AI) but it doesn't need to be. The disruptive potential of AI is enormous and can happen at great speed (simplistically, for example, stock market crashes occur in milliseconds - albeit there are some safety measures now). I have witnessed a demonstration of a human being 'internally' responding to an external question (i.e. not responding verbally or in any other physical way); the response was read electronically (by reading brainwaves coupled with - not sure how to describe it - but by sensors which were attached to the persons vocal chords (outside of the neck) and parts of the mouth which would be triggered if an attempt was made to 'speak'. The individuals 'response' was aggressive in intent. The computer on the receiving end of this enacted its AI capability and sent off various pieces of code to other machines (including AI). I find the potential of that to be troublesome. So did the researchers: to such an extent that they even amended their research to limit it to just the measuring of that 'quasi-vocalisation'. I'm not saying that humans still weren't in control of that situation, but I well remember the concerns of the inventors of the A-bomb. I agree that AI may not be aware, in our human sense, of the consequences of its actions/inactions (although see my earlier point about self-awareness) but it doesn't have to be in order for there to be consequences. On a more mundane level (but nonetheless significant for those concerned), the failure to have safeguards to protect individuals (on a moral plane), often vulnerable, from being bombarded by hateful, spiteful 'messages' (despite the difficulties of doing so, on a non-technical level), is a worrying trend. I would agree that its still a human 'fault' and responsibility, but the advent of AI adds another layer to it. There is now research to show that human capability is atrophying as AI and IT is becoming more capable - such support from AI/IT is not in itself necessarily a bad thing (I for one welcome sat. navigation!), but we do not seem to be as advanced in ensuring the adequacy of control not just of AI/IT but of those who 'own' and 'control' it. Imagine if the development of 'super weapons' had been just slightly more advanced in WW2 and in the control of fanatics. And we're good as a species in developing new fanatics every so often - for example, Mao's cultural revolution starving some 40+million of their own people, or the repressive measures involved, and now serviced by AI. I think AI can be put to very good use by humans (so many examples - diagnosis of healthy conditions, and so forth) , but I don't think we can afford to take our eye off the ball, so to speak.
Pete Gunnell
2019-05-30 09:18:15 +0000 UTCsry for the late answer. The new content pack took so much work. Yes, i agree. If some of the smartest minds we have warn of the problems that can come from AI, there certainly is a risk to it. Then again, there is a risk to everything we do. And yet we are conservative enough to limit new technologies in their power and spread to a level where they are in good hands most of the time. I think a active war that we started (so our fault, not the fault of Ai) is the largest risk, because i think someone would need to get out of his way to intentionally build a AI that would be able to completely destroy everything on this planet before it can be destroyed. It looks simple in SciFi movies, but earth is a really big place. Kind of hard to imagine a machine could destroy all of it. I think it is way way way more likely that we will limit AIs to be helpers instead of autonomous machines that are as intelligent as us. Superintelligence sounds fancy, but if you have an AI that is way smarter than any human, but all it can do is optimize the aerodynamic of cars, then there is zero risk of it doing anything evil. It wouldn't even know there is anything outside of it's task.
Olivio Sarikas
2019-05-29 21:17:31 +0000 UTCA problem with computer programming is that all its code is specialised - one minute of code calls out to another and another and so forth. So, its highly likely that one 'set' of AI machines can easily be programmed to recognise when it needs to ask a question (call out for the next section of code) to another specialised AI set, and do so at vastly greater speed than we humans. For example, AI in the form of facial recognition can determine which face it is surveilling and pass that information to another AI that can determine if that person is in an inappropriate location and/or behaving in a manner that calls the next question. That technology is already here (has been for quite a while) - though the speed and effectiveness of it is developing at a rapid rate partly due to 'self-learning' capabilities but also to the resources available to service it. Having asked the next question, AI can set any manner of 'flags' - essentially asking further AI to take next steps. Sometimes this can be seemingly trivial (though perhaps not to those being observed), and other times far from it. So, we traverse rapidly from a camera to all manner of options capable of taking steps with non-trivial impacts, flawed or otherwise. And what of democratic input let alone control - ask that of those controlling huge private technology platforms, for example. Currently in the UK there are far too many truly horrible examples of people whose entire lives have been decimated through no fault of there own, much of it thanks to arbitrary decisions (admittedly with policies initiated by humans - or at least some might still call them humans, others would debate it!) involving .....anyway, you get the general idea. As you say, risks and rewards. But are humans always going to 'win'? 51million people killed worldwide in World War 2 and I observe the current relative instability within Europe (and elsewhere, of course) and wonder......
Pete Gunnell
2019-05-27 10:19:21 +0000 UTCHi Pete, no, you absolutely right that AI includes high risks - higher probably than anything else with have invented so far. In regards to self-image of beauty, i think that we will adapt rather quick to that. After all, techniques to make things and also people more "beautiful" have rapidly increased in our lifetimes, but most of the time we had a quick "wow" effect and then got used to it rather quickly. The far bigger risk is that of war, and there is no doubt that AI will be used in Military Applications. Industrialization and Mass Transport have led to the massive wars and Holocausts of the past. So there is a real risk of AI leading to a third world war. However, technologies like the Atombomb not only ended world war 2, the risk and fear of atom bombs also lead to a global peace we have never seen before. I'm not sure it's possible to have high rewards without high risks - as sad and concerning as they are. And we can't wind back time - some nation will do it eventually. But so far we have survived it all. So let's hope we are smart enough to do the same this time. Also, and that is a very important point: AI is in almost all cases very specialized to a specific task and can do nothing outside of that, nor rewrite it's code to do anything but that, because it doesn't even realize there is anything but that. And we are still very far away from a general AI that would actually be able to do more than one specific thing, let alone rewrite itself or become self conscious. What you see are "rewriting itself" at the moment is not really that. It rather means that it adjusts it's parameters to come closer to a goal set by us. It's basically like when you adjust settings on your camera. No matter what settings you use, it's still a photocamera and not setting will turn it into a coffee machine.
Olivio Sarikas
2019-05-25 16:25:35 +0000 UTCInteresting. I do worry that AI might also damage part of our humanity; In considering the reaction that some people have to existing IT in the shape of, for example, social media - in particular, what could/should be considered as its misuse by the pressure it imposes on some and the adverse reactions (suicides etc) that can result. Its possible, of course, to argue that art/creativity has always been able to offend or indeed be susceptible to misuse; but when AI is more predominant, and especially as AI is increasing capable of 'self-programming', is it also possible that our humanity is increasing removed from such scenarios with consequent risks. Too dystopian perhaps?
Pete Gunnell
2019-05-24 10:16:26 +0000 UTC
