chippwalters
patreon
AI: the eminent danger today.
🕑 Added 2024-02-29 22:50:19 +0000 UTCHaving recently watched a video on the new and upcoming video creation AIs that will generate 30 minute documentaries from a single prompt, I thought it time I write about it.
This is a personal essay which may tell you more about who I am but also about my current thinking on the very real issues we are now facing. I believe this recent Gemini Google issue may be seen as a watershed moment for AI. Only time will tell.
----------
Many years ago, a few years after the beginning of the Internet, I was invited to give the keynote address at the South by Southwest (SXSW) Media conference in Austin Texas (back when it was smaller). The best thing about it is it's where I ended up meeting my wife.
I was then the co-founder and CEO of Human Code, a company engaged in various interactive media projects, including a game named FutureNet. This game featured Syd Mead as the concept artist and Bruce Sterling as the story writer. While I had previously worked with Syd on projects at NASA and elsewhere, meeting Bruce, a celebrated sci-fi writer and a pivotal figure in the cyberpunk genre, was a new experience for me.
At the very same SXSW conference, Bruce gave a most memorable talk. At the time, I thought it was quite the downer and attributed it to his darker predilections as a sci-fi author. In his talk he described the Internet as very dangerous. He said that every time you went online you would risk being robbed, scammed and your privacy being compromised.
This was so different from the early promise of the Internet and very hard for me and I'm sure many others to believe.
Although his outlook seemed overly pessimistic at the time, his predictions have proven remarkably prescient.
My interest in AI stems from a longstanding engagement with the field, tracing back to my work on the design of Texas Instruments' first AI workstation, the Explorer, in the early '80s.
Those machines used the languages Lisp and Prolog, which were essentially databases combined with huge chains of if-then else statements. Very similar to the original Siri.
They mostly created expert systems and had a very difficult time with any sort of general AI knowledge.
The technology pretty much stagnated until the advent of affordable parallel processors, neural networks, and deep learning, which ushered in a new era of AI innovation.
This technological advancement has been both exciting and promising, evidenced by my own experiences with AI, from purchasing a Tesla with full self-driving capabilities to assisting my wife with a project that involved creating superhero personas of people in our lives using AI and graphic design tools.
AI played a crucial role in bringing an enjoyable project to life for me. My wife has a fondness for classic video games, such as Pac-Man, prompting us to acquire a multi-game console. I designed ts exterior, for which I've shared some images. The fun part was using AI to transform family and friends into superhero figures, complete with striking resemblances. The images shown are representations of real individuals, reimagined as their superhero counterparts. This was mostly done through the use of tools like Midjourney, Inface, and Photoshop's generative fill feature, making the process both fascinating and entertaining.
I've also used AI to help me program an HTML5 canvas app to generate even-step grayscale images from a color photograph, even though I knew nothing about the HTML5 canvas API. That was impressive.
So, as far as I was concerned, this new AI promise looked pretty rosy to me.
In the past, we've had huge shifts in vocation and career opportunities based on technology. Moving from horse-drawn carriage to automobiles changed the lives of many stable owners, buggy whip manufacturers and affected all sorts of other industries.
I can remember when desktop publishing was just starting out and how threatened all of the graphic designers and layup artists were at the time. Soon they realized the simple fact they needed to relearn a vocation in order to stay relevant.
I think some of that is true for AI as well. If you look at AI as a tool and not as an end product you can see where concept designers, writers, programmers and others can use it to help them become better at their job. This is a nuanced discussion which I won't go into great detail here.
Now certainly, AI has and will continue to do a lot of good for humanity. Especially in the field of medicine and diagnostics as well as helping people learn about new subjects.
But, very much like Bruce Sterling's predictions concerning the Internet, I'm now starting to see the potential for the darker side of AI.
I think we're now starting slowly to turn the corner on the rosy outlook. Google's new Gemini has shown the world the tremendous potential for using AI to alter the perception of history and help control minds. I have now realized that even Google searches are way more biased than I had thought and only wonder how much they affect the thoughts and views of billions of people (Duckduckgo.com is certainly your friend).
AI has so permeated YouTube that it has become more and more difficult to find information on subjects you want to learn about. Just recently I was trying to do some research on a food processor and the majority of online reviews are AI generated with AI voices. Of course they are worthless and for some reason there's not a way to tell YT to ignore the channel.
This next generation of text to 30 min AI videos will flood YouTube with millions of instant documentaries on every subject imaginable, with no way to know how truthful or accurate or even interesting they are. Not to mention it will be super simple to put words into the mouths of any public figure and show them in videos doing things they shouldn't be doing. It will be very difficult anymore to actually look at video evidence and believe what you see.
People no longer read books. Now they tend to gather most all of their information from videos. This is why the new AI video technology is so alarming and threatening.
This is all happening in real time, it's not waiting for all the AIs to get together to take over the world and make us their slaves.
We do not need to wait for artificial general intelligence (AGI) to be fearful of what this technology can do. It is already poised to be the most significant propaganda weapon ever created.
Comments
MH
Welcome to the dark side Chipp, sad to have you finally on board. When you try to explain that Ai will be a net bad thing for society and not in a terminator robots killing everyone type way people relegate you to the tin hat brigade or call you a luddite. But that was always based on an underestimation of the technology and the speed it will develop. Ive been using duckduckgo for a while and ridiculed for it. Ai is: Bad for jobs. Bad for truth. Bad for trust. Good for manipulation. all you can do for the last is to stop feeding the data sets by protecting your privacy, your clicks and stop posting your whole life on SM.
Pearce
Definitely agree Chipp and Google's behaviour over the past decade has certainly ruled them out as a trustworthy not to absolutely use these tools to propagandise people. If not something entirely worse.
Chipp Walters
Yes, you couldn't be more right about search bias. Just Google white family and look in the images and you'll see half of them show diversity instead of white. For years now I have found that when I search for anything that might have any sort of bias, I always use duckduckgo. Google cannot be trusted ever since they deleted their mantra: "do no evil." And having grown up in a world where monopolies were routinely broken up by the government, I am saddened and disappointed by our current group of leaders who are only interested in taking money from Google and the big tech companies rather than actually govern the people. Google has over 95% of the search market. That is the very definition of a monopoly. Our leaders standby and do nothing.
Chipp Walters
What you say is true. AI can only get you 85 to 90% of the way there. You still can't art direct AI. Nor can you take the steering wheel out of a Tesla. And, now after the Google debacle, the overall trust in AI has taken a huge hit.
Chipp Walters
Most designers understand that everything is derivative. In fact as designers, we often create mood boards where we borrow from images that inspire us. Just look at Apple and how they've ripped off Dieter Rams. It's been going on since before AI. Still, you have a really good point about how AI will create a lifeless mishmash of art styles and overcome new artists efforts with an avalanche of images being generated everyday.
Chipp Walters
Again, as I've said. I don't have issue with the creativity process as much. What I have issue with is the specific inherent biases programmed into AIs that will overtime change minds and history. We are like the proverbial frog in boiling water.
Chipp Walters
My issue, as I stated, is not with the creative part of AI but rather its ability to propagandize and change history based on inherent biases. The fact that a single programming team can create an AI that can change the minds of billions of users is a scary thought indeed.
Chipp Walters
https://www.forbes.com/sites/cio/2024/02/29/a-litany-of-apologies-over-googles-gemini-ai/?sh=7231318dc818
Samuel
In a nutshell AI will make traditional human pieces being more valuable because making art will be easier. I have been finetunning for 2 years and a bit more right now and have more than 15 years of art projects in company works outside AI. You can't stop the progress and WE need and this is hard for some but WE NEED, WE REALLY NEED to adapt to it in these workflows OR being niche and continue to our things with the risks it may have,
Marco Chacon
What is going on with the pictures? What is specifically special about Gemini?
Sean Glenn
I started my design career just as the old school designers were loudly sounding alarms about the death of craft thanks to desktop publishing. They were right, of course. We lost all the hand lettering, all the carefully constructed grid-based layouts in books, the impeccable spacing even in the most mundane typeset copy. Ai is going to kill off a bunch of jobs, wreak havoc politically, and probably destroy the concept of artwork having value at all, and either destroy or cement copyright—not sure which is worse, honestly. My friends who work in film and television pre-production are freaking out; they either think it's all ending tomorrow or they're scrambling to figure out ways to use these tools that don't look identical to the other 2.5 million images Midjourney spits out every day. It was bad enough to get the Artstation-ification of concept and environmental design, as everyone copied the most popular and then everyone else copied the copies. Same thing happened with design thanks to Dribbble. It won't eliminate the human from the equation entirely, but the human touch is going to be something seen as a luxury. If only we had been able to build AI to wash dishes or pick strawberries, instead it's all artists and writers… It feels like such a Pandora's Box, and it's never going back in. I guess we march forwards towards Roko's Basilisk.
Keo Match
I agree on pretty much everything you’ve said although go even further and say that undoubtedly that ai has already been fully weaponised but now with the accessibility becoming even more open and user friendly it won’t be long before it could truly go nuclear (and I would like to think I’m being metaphorical here but scarily it could be a more literal statement than I’d like too). As a creative I also feel that the theft of content for ai models is absolutely disgusting but much like any kind of invasion backed by a powerful crowded swell I fear that any justice will be too little to late. Sadly I think some of the most wonderful creative minds contribution will be remix, regenerated and echoed into a meaningless void of algorithmic vomit as the same basic ideas will be repeated again and again telling the ai that these ideas are valuable because of demand and not understanding quality of substance. Will there be expressive new explorations via ai? Of course but the capacity for them to be heard and held with significance will be harder than ever and even more sparatical. What’s the answer? No idea. Adapt or die. Rebel through analogue maybe. Protect through substantiated knowledge and journalism morality while fending off the editorial and bucks based biases that are looking for your eyes and taps. I personally think graffiti in the physical space will become even more relevant and it will take on an even more important role in the digital space as plea for authenticity and independence. A future that equally light and dark but unfortunately as those 2 mix we will be coated in a dirty grey and I’m no sure we have the tools to clean these glass house windows off enough to let the light back in. One day hopefully… hopefully.
Alexander Virgint
For me, the issue with AI has always been one of labour. You touched briefly on it about relearning vocations, but these people were still experts in their fields who needed to learn how to use new tools. But AI isn't being sold as a tool, it's being sold as a replacement; for artists, for filmmakers, writers, photographers, engineers, programmers, customer service reps, every single level of the labour process is told that a machine is threatening to replace them. But a replacement would necessarily have to be competent, and AI is not competent. It makes mistakes, it lies, and it's incredibly gullible. It is only as good as the training data it has access to, and even with the incalculable wealth of information on the internet it seems only capable of producing things that "look right." This is a fundamental flaw of AI; it is an unreliable tool and a bad replacement for human labour. What it ends up being is a gift to disinformation campaigns and conmen. My expectation is the bottom of this is going to fall out eventually, but not before a lot of very large companies get roped into very expensive subscriptions and the rest of us get a completely ruined internet.
Ray L. Ackerly
And on the search thing... I saw this about 20 years ago because studies that contradicted the political narrative would stop being findable. Data just straight up deleted that I used to connect. The biggest clue was when all the news outlets starting removing comments, which previously served as a source of nuance for arguments.
Ray L. Ackerly
To be fair... the reviews issue was already happening with people. We never know who is sponsored and whatnot even at the highest levels of government worldwide. The problem to me is that money is an ai system and some inputs are pretty heavily overweighted. AI is only winning because it's competing with humans not perfection to start. And we run on toast... ;).
