fuck-ai-but-in-the-right-way
alright so lets do this thing
my position like that of all technology, is that technology works best when it's applied in the places it makes sense. i think people will generally place technology into places it's not needed or are driven by capitalism to include technology to "increase value". at the end of the day i quite like analogue machines - machines that do one specific thing and do it well, with maybe a few added quality of life features. this could go into a lengthy enshittification arguement, but that isn't the point i want to make here.
computers are just a way of processing information or instructions to achieve a goal. whether it's to produce a meaningful way of interpreting data, or to make something happening - the computer follows steps to achieve a goal. computers also have a million different ways to achieve the same thing, all of which can be subjectively argued amongst engineers and developers. the form in which computers exist today is a product of mass production and adoption - once again pushed by capitalism. the form a computer exists in today is not necessarily the most effective or perfect iteration of itself. it's just the one that's had the most money and time poured into it since it made money, and it made money being sold as well as made money performing a task.
the tasks those computers perform are supposed to just automate or simplify tasks that we did manually in the past - or open gateways to new ways of doing work that we couldn't have done in the past manually without incredible special training or education. usually these tasks weren't "art" in a sense, they just needed to be done and a way of automating them had not been achieved. processing data, while possible for the human brain, isn't entirely it's purpose.
at their most basic function, our brains are designed to survive long enough so we can procreate - that's the entire scope of our programming. however we have sentience, we can experience intense emotions about a large scope of existence - and we've developed mental concepts such as language and expression. our sentience intertwines with our primal need to survive and procreate - having money and wealth can signify to others we have stability and health to procreate well with. however the need to express yourself through creation - of art, music, visual, poetic, meaningful, and pragmatic - also shows the primal brain skills that can be useful in survival and procreation - growth and progress.
now i like stability - in every fucking rts/city builder i play i hit stability and i feel anxious about progressing further to upset that stability. however a challenge this does not make - taking risks, being encouraged by others to take risks, is where i learn to adapt and survive. my sense of nihilsim fails me here because while nothing matters to me long term - i want to survive, and i want to win. you can't want to fight to survive and be nihilistic at the same time - true nihilism means giving in to entropy and embracing the void since your existence has always been null. i however choose to seek change - because it's interesting. i like roadworks because they are doing something. i like new things because it keeps the world interesting and fresh. nihilsim would have me see these things as neutral because they don't matter - but for my existence, short as it is, they matter to me.
before i get even more off track - let me align this with my point. everything i have said so far can be summarised as thus - we are primal creatures with a god-like level of sentience - the physical and metaphysical effects we can have on our world are immense - but at the core we are driven by our primal traits. we create tools because we can, we can because we want to thrive, we want to thrive to experience the world and what it has to offer. creating tools is about creating abundance to fulfil our want and need to create new things and move forward. being sentient drives us to look inward and understand ourselves - and our senses give us tangible ways to do that. we can feel and listen and create music - we can experience colour and touch to make art, we can think and feel to create meaning.
we've crafted a world in which we must do things in order to keep the world moving - we create things to fulfil our need to have meaning - the things we create fill a need, of either survival or to show we survived and can now just sit with our existence through self expression and discovery.
the technology we build should solve a problem, or remove the need for us to be involved - in a perfect world this should free us to just exist - but right now the only people who are freed are the ones who own the solutions to problems - not even those who create the solutions are free, as they are owned by people with power and wealth. wealth should create abundance for all, instead it creates abundance for a few.
now lets bring in llms. you have a pile of maths that can be used to weigh inputs to create likely outputs. with the sheer amount of data in our world to consume and parse, llms are the perfect tools to identify patterns in datasets that are too large for our brains to comprehend. a side effect of this is that they can produce likely ouputs for ALL types of data. we digitise our data - we turn it into formats for computers to consume and parse. we've created small pocket sized machines that can digitise the world around us at a whim and turn it into digistable data for computers. this didn't happen all at once - it happened over time, solving many small problems or challenges people had to share information.
now for the purely econmical minded - who view art and visuals as data points on a spreadsheet - this means they can generate slop for people to consume - in the same way a tech bro now thinks they can create solutions for new problems with existing data. to an extent - they aren't wrong. yes, they can do both - because we've created a world in which this quantifies success. our world flourishes on short-term success, we look for short term solutions for far-reaching problems because it's cheap and easy - the quantifiers for success we've defined through our leaders and our capitalist societies. something is not valuable because it's expensive and takes a long time - it's valuable because it's cheap, easy, and has a large profit to investment ratio
this happens in every industry across the board - it's not limited just to creative industries.
however like in most cases, when you show people a new tool that "does everything" they literally try and use it for everything even when it's not the most efficient solution. llms are currently going through this phase. they solve lots of problems because they've been designed and modelled to have an answer for everything. real life isn't like they - there isn't a perfect answer for everything. if you're an expert in your field you know what the best fit answer is, not the perfect answer.
i really enjoy cross-disciplinary solutions, and i think llms could really crack open patterns across multiple fields - but at the end of the day, you want to craft a tool once and use it multiple times - not craft a tool everytime you need one. every survival game player will tell you that's wasteful.
and this is where my vision between llms being used practically vs being used to fulfil expression really takes hold. you should be able to express yourself through art without it being tied to capitalism. you're not creating art to fulfil a goal - marketing uses art to complete it's goal - but it is not an expression of feeling - it's psychological in nature. having llms produce a final output is a weird concept to me, because they are an accumluation of knowledge/patterns from the past. there's a concept in art where you can't break the rules till you know what they are. llms are limited in this capacity - they know the rules, and they follow them - because that is the pattern. the things they "discover" are a result of identifying patterns that exist across multi-form data that humans could not feasibly process over decades of personal research. this is their advantage as a tool.
llms being used in the art world rn - for enterprise and corperate goals - to make money. this fulfils a goal, but not a desire. to the people who only care about succeeding in a capitalist manner this is exciting - they already the production of art to the lowest bidder already. llms just highlight this failure to value human art and expression.
the patterns that exist in enterprise and corporate worlds only exist because that's how they evolved - marketing needs content, content requires design and art - those designs and art follow patterns. boring patterns, but patterns nevertheless. so they can be spat out of a pattern emulation machine. but again, llms are just highlighting an existing issue. simon scales at cdw talked about doing like 100 - 200 concept pieces a week for hollywood - i would say while that art had human intent and artistic vision behind it - it was commercialised and was part of a pipeline to increase productivity.
his expertise was in identifying patterns in world building and design, the same thing an llm does - it doesn't express a fully realized concept or creative choice - just an artistic stereotype to churn out art to.
having an llm do this, to me, just highlights the main core of the ickiness being how exploitive and oversaturated "arts" is as an industry - let people create, take risks with new ideas, fail with new ideas - learn from those failures - failing is being human. we didn't get this far by always being right. new patterns can only form with combinations of things that haven't existed together before. this is the illusion of creativity that llms exist behind. they combine existing patterns in new ways, but the patterns themselves aren't new.
some of the existing patterns they combine are actually terrible and inefficient, or they are emotional or redundant to us - but our sense of familiarity and comfort from the known makes us afraid of removing redundant parts of our lives from the equation. this is across the board, all fields, all expression, all industry. humans are an extremely nostalgic and rooted in rituals, routines, and culture. everything must be preserved but we must keep evolving. these two incompatible notions compete within us. our primal instints making themselves known again - what works made us who we are, but what works now was something new once upon a time.
i have many faults with the left, and a big one rn is their refusal to do anything other than whinge about llms under the hype of "ai". they sound like luddites and doomers. they should be making an effort to understand - to see where this technology is actually helpful and how it can be used ethnically, and not just outright refusing to learn anything. china's current government grew powerful on anti-intellectulism and i see this growing in the left. much to my distaste.
just becuase something can be used unethically doesn't make the very concept unethical or immoral.
i am dragging on a bit and i could continue to write about this for hours, but for now, lets conclude. llms are a tool. they fulfil a purpose based on a cumulation of existing data. it's a useful tool and a useful piece of technology. it has a place. unfortuntely it's creation (for the large amount of publically available models) is born from data that didn't belong to the creators, and they are reaping the capital benefits of that act and not passing them onto who they stole data from. however the technology isn't the reason that happened, it's the way they constructured their product using the technology without any consequence is why that happened.
the intent behind your use of an llm is important. i feel like the generative ai we see people getting upset about is it's own lane. if people are generating art because they want to use it on their website or their profile or something that is meant to express them is disingenuous to who they are. if you're not artistic, why pretend - if you like art, then pay a creative. half the time i think about designing a website i think about it being practical, not artistic. it needs to convey who i am, not who i want to be. you could probably boil down an unhealthy amount of llm usage to people projecting who they want to be through an llm. in the same way we have brainrot exploiting peoples dopamine, llms do the same thing. it's why they are addictive - like the tools of medicine that are used to heal and make impossible things possible medically - they can be abused for enjoyment or feeling good. tools used in the wrong way create harm.
there are endless sci-fi stories about a life changing technology being over-used and overly dependant upon - creatives have been thinking about this for years. it's not new - the fear and the misuse of labour and exploitation has always been there - llms just make it easier and paint a huge "i like misusing exploitive labour for capital gain" sign on yourself. as well as "the work i do is meaningless and tedious, so i want to offload it" - identify when it's a fkn problem, and make systemic change in those areas where llms highlight wasted labour and effort.
practical use of tools makes them useful, any other emphemeral purpose is meaningless in the long run, and i think you need a touch of nihilism to see that.