Main, food, history, music, tech, money, flora-fauna

ArtEfficient Intelligence

: tech, October 11 2022


Your corporation is probably blasting your ears with slogans like "digitization is the future", "automation is the key" and so on..., but only recently the management bobbleheads started using "Aritficial Intelligence" in their presentations. Of course, none of them knows exactly what this means, let alone what it involves. These people are so obsessed with their own agenda and focused on the political correctness, that you can sell them dogshit with an "AI" stamp and they will take it. But let them be. To make it a bit clearer I will try to explain what is usually meant nowadays when people say "AI". Are you ready? Here we go.
Advanced statistics.(notice the period at the end of the sentence)

Now, let's try to define what "intelligence" (including artificial) should involve. The opening article in Wikipedia describes it best.

Intelligence has been defined in many ways: the capacity for abstraction, logic, understanding, self-awareness, learning, emotional knowledge, reasoning, planning, creativity, critical thinking, and problem-solving. More generally, it can be described as the ability to perceive or infer information, and to retain it as knowledge to be applied towards adaptive behaviors within an environment or context.
It is all of the above mentioned qualities. One may argue that the machine learning somehow touches these topics, but that's not the case. ML is closer to mathematics and statistical analysis than real intelligence. Other may say your brain works the same way - no, it's not. Your brain uses layers of so called "cognitive and sensory pathways", so complex, nobody can even dare to try to describe how they function. For sure it is not the "if-then-else" construct, used in the puny exaggeration that most of you call AI. In other simple words - AI (as it exists today, in public) is not a real intelligence. It is a mathematical algorithm to analyze and interpret data. It is not some magical thing, learning and evolving past the parameters it has. For sure, it may lead to some unexpected situations, but so are the bugs in the open-world games. I know about a case where 2 AIs created their own language to communicate with one-another and this proves nothing - this was their purpose all along. After billion years of doing so, they will not stand up and walk to each other or go discover another AI, as the concepts of evolution and abstract thinking are not present in their initial matrix.
I was thinking of the "AI" implications for quite some time, but what really draw my (superficial) attention was an AI winning a drawing competition. Of course the judges had no idea what they were voting for.
An artwork made by Artificial Intelligence (AI) won first place at the Colorado State Fair’s fine arts competition last week, sparking controversy about whether AI-generated art can be used to compete in competitions. The winning artwork, submitted by Jason Allen, was created using the Midjourney program (which offers 25 free uses before requiring subscription) in combination with AI Gigapixel. The result came in first at the competition and won a $300 cash prize.
The picture "Space Opera Theater" (above) is created by Midjourney and is absolutely stunning! DALLE2 is another "hot right now" AI-painter. I checked/tested them both these days. The title picture is created by the latter. The string was "oil painting artificial intelligence explain to human it exists". Usually the bots create 4 pics. My 4th one was absolutely off topic, but the first 3 ... well, I would say they were pretty good. On the first you see a young man, teaching a robot what the letter "A" is - the robot interprets it binary. You can tell immediately that it is a machine by the sharp features and overall industrial look. On the second, a rather human-looking machine is having a conversation with a mid-aged man. The robot is cunning and feels he is really challenging the person. On the third pic a man in the background is holding his head in despair. Some really crude looking machines ... look happy. The picture lacks the focus and the distinctive lines of the first two. Do they represent some chronological timeline? The AI might have created them one after the other, or using 4 threads simultaneously...My lack of drawing expertise would say these 3 were pretty good. Reminded me a little of what an artis would draw in 1960s if asked to create a book cover for an Isaac Asimov's book.

Today I enrolled also with Midjourney and my first call was "oil painting,epic,emperor of mankind,beacon of light,cathedral,great hall,crowd,volume light,realistic,red,gold,black". The one on the left is the result (picked 1 out of 4 and refined it). If I have not seen some MJ creations before, my jaw would definitely hit the deck. Me and my wife both agreed we would be happy to have some high-quality prints on our walls as Midjourney creates FANTASTIC sceneries. Again, looking at them, you (only!) have the feeling what they represent, but ...if you concentrate on a certain part of the artwork ... you cannot really tell what it is. Some of the contours seem familiar, yet unidentifiable. Same is the case with the real modernist artists. I see no problem with that.

So is this an AI?
No, it is not. It is a really good algorithm, that combines different visual objects really well. If you take a picture of a flag, a soldier and a beach, obfuscate them using several known Photoshop filters and then splice them together, you will get more or less the same thing. To make it look good, a professional would spend several hours in the works - the bot does it in seconds. I am not saying it is not a spectacular feat of human engineering genius. It absolutely is and, in the future, when HQ-export is possible, I would definitely put some wallpapers around the house. But it is just an algorithm. If you spend long enough time, you will notice it uses the same patterns, when receiving similar input. DALLE2 does not even differentiate between "a cowboy riding horse" and "a horse riding cowboy". It also has some weird "quirks" to match the current woke narrative! (What a coincidence, huh?!)
In September 2022, OpenAI confirmed to The Verge that DALL-E invisibly inserts phrases into user prompts in order to address bias in results; for instance, "black man" and "Asian woman" are inserted into prompts that do not specify gender or race
Another funny outcome of this "AI winning a contest" story is the outrage in some art circles. Apparently these so-called artists dreaded the AI as much as others would question their skill and the value of their artwork. Some started fantasizing how in 2 years AI will produce its own movies, theatrical plays, video games. I can only pity such morons. To fear an algorithm can take your bread means you are mediocre at best, your "art" is mostly garbage and you lack creativity overall. If you are easily replaceable by a script, start thinking of another area to make a living. I experienced the same drama ~10 years ago (never been influenced by it though) when everyone in the IT started panicking over the "cloud taking over our jobs" and millions of IT profis dying of hunger. What really happened is developers are desperately sought after, cloud and security engineers basically raise their hand and get hired almost immediately. My humble prediction is that AI will be a new field for the artists to explore and collaborate. Working together will bring even more amazing stuff!

But why stop here?! Let's check other areas where the AI is the new gold rush. You guess it - the automotive industry and the "self driving cars". 13 years ago Google showcased their first (potentially) self-driving prototype. At that time I was absolutely stunned at this idea. Train drivers and maybe pilots would have been the first ones to lose their jobs, but driving a car has a magnitude more variables and unexpected situations to happen. If you ask someone from "the industry", he will explain you there is a revolution in the sector. If you ask the regular customer (like me) - nothing radically changed. Hardware became smaller and faster, programs (just maybe) a little bit better, but in its core, the tech is still a myth. Here is what a pioneer in the branch confessed 2 days ago:
George Hotz, the serial entrepreneur who founded self-driving startup Comma.ai, has this to say about autonomous cars: "It’s a scam. These companies have squandered tens of billions of dollars."
But despite all the money thrown at the problem by Google, Uber, Tesla, Zoox, Cruise, and countless others, autonomous cars are no closer to reality today than they were 5 years ago, and many of those companies are running for the exits, looking to cut their losses if they can.
He himself, a renown hacker, raised $18 million for his own driverless concept in the past 7 years. Let me tell you when/how you would know if we have achieved it - when your car insurance company agrees that all claims should go to the car's manufacturer and not to the owner of the vehicle. Only then we can really say "OK, we now have a self-driving car".
Let's step back for a second. In what context did you first encountered the idea of AI? I can bet it was either a robotics-related sci-fi book or his majesty "The Terminator (2)". And what did the AI do - yep, simple - it killed. So which organization, that existed in the entire human history, in literally every culture and society, would spare no effort or resource to get it? Yep, you are right again - the military. My best guess, derived from our newest history (world war 1 since) is, if something is commercially available for me and you, the guys in green have it for at least a decade. Meaning IF there is an AI at all, in its true meaning, the military has it. This will be something top secret, as a breakthrough like this would mean the difference the gunpowder made during the middle ages. Some military analysts predict achieving an "AI Supremacy" may trigger the next global war. That's how serious the things are. So next time your boss overhypes the new ERP software, stamping it with the AI badge, just pass by with a smile and a bit of salt, do not overthink it, it is just the next buzzword, nobody really understands it.