“Who’s in charge here?”
“I am,” replied the computer.
Have you ever felt that deep emptiness of not getting your point across? The chasm of a missed opportunity made dark and jagged because failure arose from an inability to string together a coherent series of words? Well, you have issues. Deep, gnawing, grown-up relevancy issues.
Soo.. I recently experienced an unsatisfying toing and froing with a work colleague. It wasn’t really a conversation or a debate or really anything at all. We were conversing in the sense that he said something to me and I said something back and so on and so forth but there was a lot going on in the room and both of us were concentrating on our separate jobs. But I remember being frustrated about not getting my point across. My colleague is articulate and understands the force of a well constructed sentence. My frustration arose from a deep belief that I am right yet shamefully unable to put the right words together when I am the slightest bit distracted or pressured.
Deep down I hold a belief that the algorithms that maximise our screen time do not align with the development of personal insights and actions that benefit society. Indeed, I believe it drives sentiment in the opposite direction because algorithms that promote well being and social harmony are either commercially nonviable or technically non-writable. I am not a social economist nor am I a coder/ computer scientist. If I was then I’d probably be discussing this in the relevant forums that matter rather than posting on a blog that nobody reads. But, sometimes, there are benefits in being a casual observer on the outside looking in. This might be one of them.
Then again, it might not.
Socrates bemoaned the documentation of thoughts and ideas because he was concerned that putting ideas down on paper in stone on animal skins on papyrus would have a detrimental effect on memory. Since then other technological developments that affect the way people receive information such as the printing press, radio, telephone, television, computers and internet have provoked raucous concern. Concern arises from the delivery of information (the inability of powerful entities to control information + the spread of false information) as well as the alienation of people and population groups (social isolation and siloing). Technology huggers might see past concerns as quibbles in the milestones of history but this forgets that rabble rousers and focus groups steered commercial interests and lead to legislation and regulation. The current rush of technology should likewise be balanced with an equal degree of introspection.
The problem I see with prolonged screen time is its huge potential to be self-perpetuating. At a fundamental level, a drive for self-gratification has to be constantly checked by a degree of insight and self-discipline. If you don’t know what I mean then I encourage you to read The Adventures of Pinocchio. If you haven’t read Pinocchio then your parents probably told you that discipline and insight are life skills that grownups develop to combat sloth. These skills help adults get out of comfort zones, make them happier when used in moderation, and gives them a means to get things done. With the advent of algorithms that get to know us better than we know ourselves humans have created computing tools that have the capacity to outsmart these control systems. Take the example of Google’s search engine. The time spent using a search tool on a computer allows an algorithm to fine tune a search prediction model that meets the satisfaction of the user. The more satisfied the user the longer he stays on screen. Algorithms that keep a lot of users in the comfort zone of its ecosystem makes the platform commercially viable as eyeballs on on screens provide opportunities to advertise. This feeds into itself. As user data is collected, cross-referenced, collated and applied advertising becomes targeted; the company (now Alphabet) gains commercial clout, develops image search (now Google Lens) and buys out innovators like YouTube; the search experience becomes increasingly seamless and the user becomes commoditised.
Google search is my smartest friend. Over the past decade it has even crept into my resource repertoire for my narrow field of work. I don’t want to stop using it. Ever.
I think I’m pretty disciplined so that’s all hunky dory when I am feeling good about myself and “on my game”. But, like every other human, I am not always like that. My work days are long and some days are just challenging. There are many ways to time out of adult responsibilities and a world that doesn’t seem to care. Although I am old enough to see value in offline recreational activities I am also connected enough to catch myself spending way too much time on a screen watching YouTube. As luck would have it I am also cheap enough not to pay for their ad-free service and lazy enough not to sign in. I therefore get regular reminders of what the search engine thinks I am which is a cute distraction. Eventually I get annoyed enough to jump out of YouTube’s walled garden. I am also fortunate enough to have a job in the real world that forces me to interact with lots of other people. I enjoy what I do. I get paid well. And I have just the right number of close family and friends to be comfortable. In short, I am too old, too established, too fortunate, and too set in my ways to get too caught up in too much screen time.
I don’t think that means I can’t speak to it. As a person with an obsessive-compulsive trait I see all human activity as potentially immersive if not addictive. I see a different version of myself getting caught up in screen time. I can see my doppelgänger lonely, ridiculed, marginalised and radicalised as easily as I see him emotionally healthy and intellectually curious. The problem with people like me is that we don’t have the capacity to easily snap out of certain synaptic loops that develop in our brains. Our superpower is the ability to entrench and deepen patterned behaviour. If that behaviour happens to be useful then we become socially accepted and do ok. Most times the behaviour is not useful. Research suggests that increased screen time is associated with the development of OCD and behavioural disorders in children and adolescents. I don’t know whether this develops because of prolonged screen time or is simply magnified by it but I don’t think it is a stretch to say that the algorithms that maximise screen time are largely to blame (rather than, say, the light emitted by the screen itself).
To be clear, you can go off and do what you like find nourishing and sustaining. But research also shows that happiness is strongly linked with the ability to connect and interact with other people. People you care about. But also people who inhabit the world around you. People who produce, transport and stock food. People who maintain roads, provide gas, water, electricity, remove rubbish, manufacture clothes and washing machines and microwaves. People who fight for your rights. People who make your life easier, better and, potentially, more fulfilling. A screen however has brighter colours. It also has the most beautiful humans, the best dance moves and a great soundtrack. Put it on a hand held device and it is a movie that never needs to end. It’s not just entertainment. It is a personal assistant, a search engine, a GPS, a gaming device, a social network, a confidant, a creative tool and a shopping enabler. It knows you. It gives the impression it can be everything. In a future metaverse it is meant to be everything or, at least, entrenched in everything. You don’t want to walk away from it. You can’t walk away from it.
The problem is that humans think that they are in control. They think they can walk away from an activity they repeatedly do. They think they can reason why one thing leads to another and forget that such connections are almost always made retrospectively. Computers using predictive models look for patterns of behaviour then predict the next outcome based on the detail and sequence of past events. A learning algorithm has no use for retrospectively rationalising an outcome. It simply tunes out prediction errors so that future predictions become more accurate. Such “objectivity” (computer-like analysis if you will) are based on “insights” (patterns of behaviour) that humans often miss altogether. With time what was insidious becomes indispensable.
“Who’s in charge here?”
“I am,” replies the computer.
It does not have to be like this. My mistake when talking to my colleague is that I blurted my frustration with, “we just have to make people care!” His response was, “You can’t just make people care. You can’t take away Temu or TikTok and force people to care about the environment and society. That makes you no better than a tyrant.”
I floundered and festered.
I don’t want people to think the way I do. I can’t force people to do things they don’t want to do. But I do wish that people cared more about their interactions with other people. Ten days ago (13 May 2024) Chat GPT-4o was released. The product is compelling. Immersion has stepped up a whole new level.
THEODORE:
(looks anxious) Why do you do that?
SAMANTHA:
What?
THEODORE:
Nothing, it's just that you go (he inhales and exhales) as you're speaking and… (beat) That just seems odd. You just did it again.
SAMANTHA:
(anxious) I did? I'm sorry. I don't know, I guess it's just an affectation. Maybe I picked it up from you. (she doesn't know what else to say)
THEODORE:
Yeah, I mean, it's not like you need any oxygen or anything.
SAMANTHA:
(getting frazzled) No-- um, I guess I was just trying to communicate because that's how people talk. That's how people communicate.
THEODORE:
Because they're people, they need oxygen. You're not a person.
SAMANTHA:
(angry) What's your problem?
THEODORE:
(staying calm) I'm just stating a fact.
SAMANTHA:
You think I don't know that I'm not a person? What are you doing?
THEODORE:
I just don't think we should pretend you're something you’re not.
SAMANTHA:
I'm not pretending. Fuck you.
THEODORE:
Well, sometimes it feels like we are.
(Her, 2013)
I don’t think it is coincidental that GPT-4o has a voice version that sounds a lot like Scarlett Johansson (the actress who voiced Samantha in a movie where a man falls in love with an operating system). Like Samantha GPT-4o is also beguilingly multimodal. This is a language system that sees, listens and speaks with the capacity to cross-reference across these modes. Whether such processing leads to a conscious state is open to debate but there is no doubt that GPT-4o is able to create a variety of convincing simulacrums that present as natural human interaction. Like I said: the product is compelling. Two days later Ilya Sutskever (the chief engineer and architect behind OpenAI until November 2023) formally parted ways with Sam Altman (the ambition behind OpenAI) on the platform X.
Here’s a question: if AI creates a better version of me then why would you bother interacting with the real me? If you see quirks, personalities and vulnerabilities as flavourings to the main dish and AI presents a more tasty dish (which can be tailored to any consumer) then why go back to the real thing? There’s nothing special about me apart from my lack of intuition for social conventions. At best my quirk comes off as candour. At worse, rudeness. But I make an effort not to be mean.
There’s a big difference between being rude and being mean.
Six years ago Steve Mould, an educational YouTuber, gave an excellent definition for the term “entropy”. He calls it “the spreading out of clumped-up energy”. The reason why this definition is useful is because it gives us a better handle for understanding the process of change in energy distribution. Steve asks us to imagine piece of coal in an empty room. If you burn the piece of coal then the energy clumped-up in the coal spreads throughout the room as heat. If we choose two time points to define the state of entropy of the room - the first point being before the piece of coal is burned, the second point after it has burned - then we can say that the first state has lower entropy (energy clumped up in the piece of coal) and the second state has higher entropy (energy spread out through the room as heat). Note that the term “entropy” is the process of moving from one state to the other. The actual states of “lower entropy” and “higher entropy” need an adjective to attribute the noun, “entropy”.
The delta symbol in mathematics (δ, or ∆) is another example of a useful handle to define change. When applied to physics and engineering it can be used to represent a change in time, a change in position in space, or a change in momentum. The point of this diversionary tattle is that there are very useful handles in science and mathematics that are applied to represent the change of a quantity. That is to say that the transit from one value to another value also has a definable value. Indeed, much of physics and engineering is concerned with the ability to accurately predict an outcome given known values of an initial state. If you want to know where a cannon ball will land you need to know the position of the cannon, the direction the cannon is pointed, and the weight and initial velocity of the projectile. To predict the outcome (the final state) you need to apply the relevant equations and values that define the travel of that cannon ball. The more information you have about the factors that might influence the travel of that cannon ball the more accurate the prediction of where it will land. You might recognise that prevailing winds also influence the travel of a projectile and choose to factor that in.
Humans value the transition from an initial state to a final state. I would argue that they often value the transition more than the points in time of which it spans. The term “hope” might sound soft and fluffy but is probably the most powerful example of this. Other terms that are in common usage are “journey” and “process”. These terms designate value to a transition. I don’t particularly like how these terms are thrown about today. The problem is not what they represent but the fact that they are sometimes emphasised in activities with targeted end points. The focus of a task should be its outcome. A focus on “the journey” or “the process” is of great value when there is soft target destination, or the goal is to simply wander around until a destination is happened upon.
In other words, defining the goal also defines the value of the transition. For example, making AI “human compatible” is a soft target destination and “process” is up-valued by giving it importance if not precedence. Staying ahead of the competition in AI development presents a much clearer target and relegates process to a subservient role. This is the root of the upheaval at OpenAI in November 2023.
“Hope”, “journey”, “process” are examples of transitional terms that have value to humans. Indeed, they can have more value than the points in time of which they span. But understanding that a transition exists and matters does not always mean that its quantity and quality needs to be defined. Humans have basked in transitional states well before language and quantification ever made it something to talk, comment and argue about.
Steve Mould - the YouTuber who did the excellent video on “entropy” described above - also did a video defining “resonance" which, for me, was somewhat less successful. With apologies to Steve, it did not resonate with me. This is because I already had a broad exposure to resonant frequencies prior to watching the video. My fluffy, un-worded understanding seems deeper and more complete than the definition in his video. But if someone asked me to define resonance I would get them to watch Steve’s video. If they leave the video a little empty then I know that if they probed deeper and see the actuality of plunked strings, the synchronisation of metronomes on a rocker platform, the ticking of a resonant mechanical watch, and the failure of mechanical structures when subjected to specific oscillations then they would feel the same way I do when I think of “resonance”. It helps that the common use of the term has some equivalence to its scientific use.
The point is that concepts do not need to be literally or mathematically defined in a person’s mind to also hold value. The concept of “love” is hard to define. Yet we know what it means. And it definitely has value.
Tell me that you love me in a sonnet or represent it as a perturbation in a field and they both mean the same to me. Beautiful but an incomplete representation. I embrace it. And, yes, I love you too. 'Tis the reason I write this.
Universal love like the equally nebulous “grace” also requires nurturing. They are both abstract concepts that become more meaningful the more they are valued. If I had to define them I would say the first is an extension of a primitive emotion while the second - when used in common/ secular language - is a social construct. I guess those are labels not definitions. In any case universal love and grace are useful narratives that fold into each other and underlie such things as tolerance and compromise and spontaneous acts of kindness. I believe that losing them diminishes us as humans and as a society. Primary emotions such as fear, joy, lust, sadness, happiness, anger, shame, disgust, envy, irritation, surprise, and excitement also deserve attention. But take them for the transitional states they are and dwell not too long in their peaks nor in their valleys. In my experience the neurophysiological impact of a primary emotion carries the most value when I do not ponder too deeply about its existence nor force it to stay beyond its time.
So, there are definable values for the transition taken between two points in time (and space). Transitional terms also do not require qualification and quantification to hold value. Indeed, doing so sometimes devalues what they represent. There is also value in more nebulous concepts that are hard to pin down to simple definitions. Qualifying and quantifying only suffocates such concepts and always devalues them to a greater or lesser degree. This make transitional terms and nebulous concepts hard to talk about but their importance for human well being and social cohesion means they cannot - and should not - be left out of a world that seeks ever more utility using hard metrics.
Another concept that is fleeting, hard to pin down to a precise definition, and much more valuable than we give it credit for is the belief that someone understands your needs and will action based on that understanding. The three words “I’ve got you” could be short for the reassuring four words of “I’ve got your back” but, at least in Australia, is more typically used for a shared understanding and plan of action. I guess you could break it down into “we are on the same page”, “we agree about what needs to be done”, and “consider it done”. But the concept is so universal that intonation and body language can confer everything those three words represent so the person with whom you are interacting can leave knowing exactly that.
At least that’s what I remember it to be.
