AI- it is evolving

 Only, stupidly

The last time I delved into the innerworkings of AI it was in a speculative degree to assess whether or not the AI advent horizon was looming down upon us or merely a speculation for this lifetime. We took a generous look at learning algorithms and AI humanity tests and from that saw a array of the places that AI was and where it was headed. But since then the conversation around what AI can do, as well as it's availability to the general public, has all but skyrocketed. You merely have to search about on Google a bit to see countless of diatribes on the dangers of AI influence on our creative careers, the reliance of AI in customer support capacities and the utter uselessness of current AI platforms to verify any information, despite the increasing reliance on that very function. Go on Bing, if you dare, and you'll quite literally be prompted to use an AI in order to perform whatever search you were interested in. Clearly a line in the sand has been crossed.

Midjourney and Image generation was probably the start of the public interest in this new wave of artificial sensationalism. Seeing an algorithm take a single string of prompt text and turn that into a visually appealing image based on those speculations felt like a touch of magic, and a kick in the face to the many artists who's work was broken down and reworked in order to create those images. You see, despite the blow of it- AI still very much works on more complex versions of that same learning model we looked at last time. It takes existing stimuli, programmed into it, and attempts to rework the elements of that stimuli within it's operating parameters in countless different ways to fit certain 'success parameters'. Such parameters are typically also based on existing pieces of work, meaning that AI cannot fundamentally create something that is distinctly different from what already exists. So right away we can debunk those saying that 'true AI is here'! No it isn't, and I don't know of a single AI leader who even has an inkling of how to start working on it as of yet.

For the time being you could leave AI to teach itself for 100 years without any interaction from the human world and end up with an AI taste in art that hasn't evolved during all of that time, because there is no way for AI to generate new ideas as of yet. In fact, AI is so bad at new ideas that it's composites almost always carry with them a horribly disfigured signature from the artist who's work it's stolen. I always thought signing art was weird, but it seems I was always the blind one because that's exactly what is catching all of these learning algorithms off. But you can see why this is still rubbing artists the wrong way. If a machine can, pretty inexpensively, replicate their talents- then who's to say their talents will still be of value several years from now? Without regulation, it would seem nothing.

Of course, AI doesn't just stop at art- it's coming for the written word too. Language models like ChatGPT do much the same thing that the other AI algorithms are doing, only they seemed to have been integrated into our daily lives much quicker and much more seamlessly than the art did. These platforms are capable of taking example articles for form structure and language and put together coherent output articles on specific topics when prompted. This has already led to reputable news outlets making articles with ChatGPT, (With only a tiny faded disclaimer at the very bottom of the article disclosing that fact) companies already setting up corporate emails to be dispensed through ChatGPT and as I've already mentioned, Bing's desire to replace it's search engines with ChatGPT.

But with this adoption has come it's own fair share of issues. For one, Bing's search engine helper has been discovered to be a little... unreliable. That is to say, if you ask the helper to read out information to you on a topic, there's a good chance it will make up stuff and lie- which shouldn't really be a surprise considering the machine is seeding it's information from an internet rife with half-truths and fabrications. Then there are searches that veer wildly from the path of simple conversation and turn oddly personal. We've seen examples of Bing's AI threatening people who challenge it, confessing love to people who chat with it for too long and just shutting down completely if it doesn't know exactly how to answer. Which should be good news for people in my shoes, I ain't completely out of a job yet. Oh wait, this isn't even my job! Why am I even invested?

Yet despite these very apparent and prevalent hang-ups, we already see people trying to take advantage of the tools as they appear. Just browse the news from the past few weeks and you'll pop across dozens of tales about students trying to get out of writing assignments by getting ChatGPT to put an essay together instead, or getting around sending a tough personal email commiserating rough times with ChatGPT's emotionless assistance. And as someone who studied English, I understand. Sometimes you come back to the same Essay everyday for a week and haven't added a single sentence. But if you don't reach that wall where you just break down and stream of conscious forth until the entire page is full of trite, then why have you even signed up for college? It's not there to beat, it's there to break you down and make you feel like insignificant crap on the way out. ChatGPT robs you of that humbling.  

Headway has been made with the democratisation of AI accessibility, even if the functionality of said AI has been rocky. This does mean that when the time comes that AI starts getting better, and adaptive learning starts becoming possible, that's going to be the sort of tools that hobbyists will be able to play around with their bedroom. From this light, I could see a future where nothing in the human imagination can possibly be constrained because of the limitless potential at their finger-tips. Of course, by that very same merit I can't help but think of the countless cliché warnings that Science Fiction has blasted us with over the years. Skynet. The Matrix. You know how it goes. Limitless power, limitless danger, Christian Bale going nuts at a stage hand- the dark future.

Curiosity bides my attention, as it always does; a yapping hound dog lapping at my cheek to watch what next feat of daring it performs. But I am a cautious observer. Dubious and careful. As much as I want to see what AI is capable of, I don't want to see it's improvement erase entire industries of creative talent, as it is already in the midst of doing for labour. The combination of machines into the work place is a matter of consternation to be chattered over for hours, and more important minds than mine have pondered it's intricacies. But even then I won't lie about being tempted for what AI might be capable of in a year, or two. What we see today already strains what I thought possible, tomorrow might break my mind all over again.