Racing to see who can shoot themselves in the foot first... On Ai.

[circuitbored]

I haven't written much here about what is referred to as "Ai", I've mostly written about many other threats to our technology & policy-driven societal decline for a reason.

It's because those things, prior to Ai, threatened our existence, harmony, & success most in the past.

They've apparently all have taken a back seat to the recent threats to our existence being raised & amplified, by Ai company employees & even company CEOs themselves, and now major news outlets have picked up the discussions by letting VERY un-technical discussions be broadcast, in order to sensationalize & weaponize this new fear in the normal daily news cue of "How will the world end part #3296"... It's bewildering that any serious company exec would say a product they actively work on could "destroy human life as we know it", or even hurt people, and still have dinner at home on the same evening, but here we are now...

I'm not a pessimist and also not an optimist... I'd rather consider myself a "practical realist", yet across the Internet and on television lately, I've been seeing tons of undercover doom & gloom stories over the "current Ai" discussion... I've also seen many friends laid off from jobs in lieu of baffling capability promises made by "Ai" marketing, and a lot of traditional job employees being laid off...

Before we proceed, I feel it's important to recap that I'm definitely not an enthusiast, fan, nor evangelist of this current iteration that is referred to nor marketed as "Ai" -- But I work very hard to be honest. open minded, and fair concerning my actual opinions & words concerning it.

At only 4 years old or less, Ai solutions seem to have infected almost every single technology discussion worldwide, yet far more "shrouded in mystery!" is involved in presenting "Ai" than favorable use cases where it actually (helps humans). We are far more likely to see clunky, overhyped, and quite questionable videos of CGI robots, talking heads making wild claims to move markets, & colorful data-flow animations, rather than seeing actual real-world product demos that really paint a picture of how & why we need these types of AI solutions in our lives, and how paying monthly for them will also likely end horribly unfavorable over time for us all if it's based on the same kind of pricing models of Comcast and/or Netflix.

In the past, we lived in social communities, where we had wise & accountable elders that answered questions and resolved conflicts through arbitration... This could probably be referred to as non-artificial intelligence, but somehow modern tech marketing has discarded respect for human knowledge & experience, and also wants to declare it a worthless, especially when you look into the ways they are claiming it as their own without paying to harvest it.

It should be no surprise that corporations are working to absolve themselves from any accountability in this new form of technology service they've developed as well, letting huge profitable corporations off the hook for both IP Theft and Accountability will possibly be more of a disaster & doomsday scenario than any "Ai gone rogue" scenario one can imagine...

Imagine a self driving car, powered by this current form of "AI" hits a pedestrian? Who should hold responsibility if the self driving car is found to be responsible? The owner? Or the company that made software that drives the car? Somehow logic& precedent both get lost when "Ai" companies can simply say they have no control over these tools and what they do.

The actual technical truth is though, that as of right now, actual humans are running every aspect of Ai behind the scenes, from training updates, to creating prompts to feed it, to maintaining & curating data sources, to development scripting & updates behind these systems... Ai is not developing and replicating itself, contrary to what they'd have everyone believe... It's heavily dependent on human brains & hands to continue operating.

It's like a bad dog owner, living in your neighborhood, when you have toddlers, that constantly leaves their dogs hungry (off leash) to roam freely, and then claiming they're not at fault.

Currently (In a nutshell), The design of these "Ai" systems are largely based upon humans "training" systems on public data (Previously made by humans) in order to operate. The design of these systems (although no one beyond each company developing it can see actual schematic designs) is often based around scripting that mimics human interpretation of large-scale data sources of formatted & unformatted data. It is reported that the development process for the most popular forms of "Ai" right now didn't start with attempts to develop a humanoid "brain", but more so that they started with scripting that performs around a logic of matching keywords and building responses based on a pre-designed "probability scale" designed to mimic human conversation and reasoning...

There are huge problems if this is a design for tools used in settings with trust and accuracy as a key basis. Almost anyone who has used free public, and even paid model variants of Ai has seen that results are often unpredictable, inaccurate, and even untrustworthy on too many occasions to be considered a fully stable & secure mission-critical management solution... In most cases, the considered use for systems of this kind are not to put them in active use in every single possible role right away, but somehow, that's exactly what many companies are doing, and often finding out the hard way that these solutions are not quite as "street ready" as marketers promised they were.

After tech layoffs surged this year, I am beginning to see a lot of low-key requests for human development jobs online beginning to re-emerge... Possibly proving that the transition into Ai didn't quite work as expected... The practicalist within me is also fighting the idea that many of these developers who are hired in roles may well end up being credited as "Ai" workers rather than as human employees, which is also far too common within the IT industry now a days too.

We're at a very important crossroads in technology, where we need to each decide what we will accept and adopt as truth, because if we don't, the decisions are being made for us, and executed upon us all that will change the landscape of our daily life significantly. If "Ai" tools are (behind the scenes) making our lives more complex and difficult, that's maybe not really a viable business case we should be fully in support of, and perhaps the rapid rate of adoption and deployment needs more actual testing, more specific & meaningful use case justifications, and also slower adoption.

All that being said, This iteration of "Ai", and I put "Ai" in quotes because there has not yet been proven to be an artificial and autonomous "brain" behind the scenes running it (more likely humans developing it rather than itself) does have practical uses...

1. It's good at math & more complex breakdowns of mathematical reasoning processes, provided the questions have been documented prior by humans.

2. It's reasonably good at cartography, this has a threat though in disclaimer -- by being subject to the political bias of the humans that develop it.

3. It's questionably good at handling conversational technical support to humans, provided no actual administrative support is necessary.

4. It's great for churning out fast basic art, rather impressive math-based art, and somewhat convenient for generating basic or mathematical music that lasts for short-use projects, only if it has prior work of humans it can work based upon, that can often spiral into copyright & a host of other legal issues.

See the pattern there? This discussion involves so many facets of controversy & complication, it may well already be in jeopardy, just on it's issue cue alone...

The future of this "Ai's" true useability is deeply hinged in compromise... The compromise of expectations on it's output, compromise in terms of the people and their work it samples, compromise in terms of where output can authentically be used, compromise in terms of credibility of those that use it in derived works, and compromise in many other aspects, some insurmountable without deviating from logical reasoning, compromises in compromises on accountability, and compromised fairness in contrast with how the human world has worked for decades prior to 4 years ago when "AI" tech was first deployed for wide-scale public use...

In many ways, these issues & compromises may serve to be it's undoing well before discussing critical technical design flaws & limitations within this current iteration of "Ai". The mention of it replacing and "unemploying" humans is a poorly though out marketing campaign, as it is so heavily dependent on humans to maintain itself, and dependent upon humans to financially sustain it... What a giant marketing mistake they made in saying that this iteration "would replace humans & take everyone's jobs"... They better hope it doesn't do that instead, or developing it further is toast!

It's an interesting discussion indeed, that should get more respect than glossed-over conference sound byte speeches by the CEOs of companies that produce these tools, and needs more knowledgeable technical discussions well beyond non-technical panels of financial investment experts, news reporters, & marketers on news-related TV shows. Put these models on game shows, show them authentically doing human tasks, demonstrate completing a full conversation with them in broad daylight... They don't do public demos (off the rails/leash) right now, because it would highlight how much users currently need to "compromise upon" in order to keep the hype alive.

In my closing opinion, the need for skilled & experienced humans will never decline, it can only artificially & disingenuously be made to look like the need for skilled humans is declining, especially if their prior work is truly what is used to "train" modern-day "Ai"/"Superintelligence" tools. If humans are forced out of skilled jobs & cast aside there will, as a natural result, be a hard stop in intelligence of these "Ai" tools by nature the minute that skilled and experienced humans fade in trades, art, science, business (etc.) completely stop posting their work online in the usual accessible places where "Ai" or LLM resources can "find and train upon" (AKA "steal" or "lift") their work from.

Feel free to connect with me on BlueSky app, you can also send feedback there & business or event requests: https://bsky.app/profile/ruffandtuff.bsky.social