Posts from Nesslabs

I subscribed to Anne-Laure Le Cunff ‘s blog a while ago but I could not find a chance to read her writings. Before diving in, there are other nice articles by Anne that I have read this week:

• Brain-training games are BS (I kind of guessed it…)

• How stress and anxiety impact your ability to focus

  She talks about attentional biases towards negative stimuli when anxious. We seem to a) try to find threats (in an excessive manner) and b) undermine other stimuli. I plan to read more about this but just knowing that should be helpful.

• Building your mental gym

  I think it is a good compilation on mental health and agility. It has 4 pillars: curiosity, creativity, mindfulness, and rest.

There seems to be an interesting claim of a study that show learning can be beneficial for stress and recovery. I checked it out though, the link given Cope with Stress, Try Learning Something New refers to the study The neurological basis of occupation . I am not entirely convinced. I tend to approach defensively to those kind of studies because it seems like favoring capitalist productivity traps — market even “sells” resting as ‘a way to perform better later’ —. I always feel more energised when I am engaged with what I am passionate about — involves learning most of the time — though so I checked it out. Nevertheless, the claims like “People who do X have less dementia” or “Activity Y correlates with better health” does not indicate causation but rather a correlation. Instead, people who read more may have better socioeconomic status, thus/or already be healthier, thus/or have higher education. So I think there is over-generalisation and weak correlation.

Another study The Impact of Sustained Engagement on Cognitive Function in Older Adults: The Synapse Project looks like it shows persistently getting involved with hard and new activities enhances memory function in older adulthood. Interestingly, they found social activities does not benefit cognitively as much as it is anticipated.

Another interesting suggestion is expressive writing which seems to be beneficial for moderating emotions . I didn’t read the study, but I generally find writing pretty beneficial even though it is not easy to observe immediately since the benefits are generally delayed.

She also links the study Study shows how taking short breaks may help our brains learn new skills which I remember hearing in a podcast (I hope it is not Huberman, he is not welcome here…). The actual study is Consolidation of human skill linked to waking hippocampo-neocortical replay . There seems to be parts to be skeptical here too, I will read more about it and write it next week.

An interesting fact she dropped: it’s estimated there is between $400 million and $1.3 billion spent on unused gym memberships in the U.S. only .

Ethology, on instinct, development and learning, robustness and plasticity

This week’s spiral started with her post The Gut Decision Matrix: When to Trust Instinct and Intuition . Even though it was an interesting post, the juicy part was the papers she gave links to.

Development evolving:The origins and meanings of instinct was a magnificent piece blurring lines of what we actually mean by ‘instinct’ — as present at birth, hardwired and/or genetically determined quality — and what it actually is or works.

There are some intrinsic tendencies of this definition, specifically instinct being:

• developed before it is used
• unchanged once developed
• shared by all members of a species.

It is trivial to know about DNA not being the primary driver of traits if you know about epigenetics. It still shows me that what seems to be happening is a lot more subtle than I thought. It seems instincts are not preprogrammed, hardwired, or genetically determined; they emerge each generation through a complex cascade of physical and biological influences instead. One of the examples given was Gilbert Gottlieb’s work ^[1] on how newly hatched chicks and ducklings are attracted to their mother’s call. It looks like they already know how to discern the voice of their mother’s from others’ and have the ‘momentum’ to follow it. This seems like a very concrete example of a hardwired behavior. Gottlieb however questioned if embryos – while in the egg – can acquire certain experiences that can shape this behavior instead of it being preprogrammed. He discovered that vocalization shapes development of auditory systems of embryos. He even showed that he can make hatchlings prefer another species’ maternal calls by changing their embryonic experiences!

Another interesting work ^[2] is on pups and their reflexes of flipping over when they are immersed in water. They sent pregnant rats to space via NASA Space Shuttle to see if gravity has an effect on the development of their vestibular system ^ℹ located in inner ear, the main sensory system for balance, spatial orientation and head rotation and acceleration. The pregnant rats are returned to Earth before 2 weeks of their delivery. Compared to the control group, astronaut offsprings failed to flip over when immersed in water. It seems under micro-gravity their vestibular system developed differently. They observed neuro-anatomic changes as well. I have been told there is another dimension to it in a cellular level but I didn’t have a chance to read that yet. Luckily, after spending another week on Earth, pups regained their righting responses. This gives rise to a discussion on another topic – robustness and plasticity – whether those changes are permanent, and constrasting to permanance, how plastic is the way this occurs. (I read about it as well, will be mentioned later on).

It is not limited to the environment where the embryo is developing. A question I had was if it was because these were interferences to earlier development or not. It turns out, it was not. It still seems earlier development can give birth to more permanent changes as well, the earlier it is, greater the chance. Faith – a cute puppy with dysfunctional forelimbs – learned to walk upright on her hind limbs. What’s even more interesting is that Faith’s body grew in a way -curved spine shifting forward her center of mass- that makes upright walking possible.

So we should rethink the dichotomy of nature vs. nurture and how we define inheritance. It seems that we cannot leave environment out. Limiting inheritance to DNA and cytoplasmic factors in the egg alone falls short. I question if this dichotomy is still useful or not, because where we draw the line between what is inherited is not easy to decide.

Next, a question arisen by pups’ ability to regain reflexes after a week was whether the changes were robust -persistent to changes in the environment- or plastic -changes with environment-. The paper Plasticity and robustness in development and evolution compiles what the literature has on this topic.

The first realisation is that plasticity and robustness are intertwined. For example if a kidney fails to form, the second kidney undergoes hypertrophy (plasticity) to compensate and it is stable. In social insects such as bees, how the larvae is are nourished changes anatomy, physiology and behavior for the rest of the life. This determines what the life will be devoted to, reproduction or caring for the nest. Temperature changes sexes of certain reptiles which is irreversible. All of those, without plasticity, would not be possible. During pregnancy, if the mother goes through stress -nutritional for example- fetus responds with reduced growth. This adaptive trade-off increases the chances of survival in the short term even though being born small may handicap long term survival. Nevertheless, the adaptation gives a chance to survive and reproduce instead of dying in the womb. All of those shed light of the greater impact of the earlier development on the organism.

Epigenetics

Through epigenetics ^ℹ the mechanism of persisting adaptive changes without changing nucleotide sequence of the DNA across mitotic cell division both stable and dynamic changes can be achieved in a cellular level. Methylation at Cytosine, Guanine and the associated histone ^ℹ a molecule around which DNA is wrapped around changes the accessibility of transcription factors for the chromatin ^ℹ condensed structure made up of DNA, RNA and proteins forming the chrosome. In other words, producing proteins from methylated regions are not possible anymore, they are silenced. It looks like non-coding RNAs guide those molecular processes. Thus, not only the difference of gene sequences, but also having changed expressions of the identical genotypes determine the change in phenotypes. It is not only adapting to environment per se, all cells in our bodies have identical genes, yet we see specialisation. A muscle cell and an eye cell has completely different phenotypes. Differentiation is achieved by expressing particular genes for each type of cell in response to signals from surrounding cells and the environment it is in, and suppressing some of them. It basically locks the parts that are not wanted to be expresssed. Silencing in early stages is persisted after each cell division. This way, stable cell differentiation enables evolution of multi-cellular organisms a.k.a. epigenesis. An hypothesis for how gene regulation first took place is silencing viral infections on the genome.

This is also important as an adaptability driver for passing down (through meiosis) plastic changes if the costs of doing so is lower. In other words, individuals with the subsequent markers and gene reorganizations can enable developing the same phenotype with a lower cost. It does not have to be direct, changes during pregnancy can induce epigenetic changes in the offspring. It seems like epigenetic mechanisms is the first step to protect better-adaptations until it became permanent with a mutation. It is like having trials on adaptations before fixating them by biasing the mutation. It is shown that methylated Cytosines are more likely to mutate to Thymine. This explains why CG sequences are under-represented in the genome.

The point taken is that development for traits require input from the environment at specific times. The character of the input — as in the pup’s visit to space case — determines the course of developing those traits. Interestingly not all changes — even if they are big — result with an effect, and also the manifestation of the effect does not have to be immediate.

The final stop was the paper The acquisition of an appetite . Navigating to the water source and drinking it when thirsty seems to have 2 phases: the appetite — the act of seeking and approaching the water — and consumption — the act of drinking the water —. They show that consumption seems to not need learning (maybe it does, who knows) but the appetite does. They use pups to demonstrate that the appetite — seeking water — is learnt during infancy. They interrupted the postnatal learning experience necessary to develop appetite and show that when prevented, appetite toward water when dehydrated lacks. They also show that by having only one pairing for a dehydration case — dehyration followed by approaching and drinking water (naturally, when pups are left with water sources) — establishes the appetitive response. I wonder if acquiring is order of magnitude easier — maybe through a tendencies like seeing patterns on toasts —.

There is also a discussion whether this learning process is a simple Pavlovian conditioning or dehydration being a modulator for conditioning, causing water or the consumption when dehydrated becomes more effective of a signal, amplifying the strength for evoking consummatory responses and reinforce water consumption.

To sum up, it seems learning plays a crucial role for developing appetite and motivation. It is rather intriguing to see that even our appetites are learnt.

That’s all for this week for ethology… There are other interesting paper that I have found that are complimentary and extensional to what I have read so far.

Next is the paper Getting a bigger slice of the pie. Effects on eating and emotion in restrained and unrestrained eaters demonstrating “what the hell effect”. When you are a restrained-eater ^ℹ meaning that you are implementing some sort of diet that restricts what you eat if you believe that you over-shoot with what you ate (in the paper they induced this feeling by playing with perceived portion sizes, so it is purely perceptive), you continue violating your diet by saying “what the hell” — as in the name of the effect —. You think like “I already messed it up, so I can continue”, you already passed the line so it does not matter how far you go which is a fallacy. You know what they say, instead you should have say “better lose the saddle than the horse”. There is another interesting point in the study. Restricted eaters who perceived that they have a smaller slice of pizza felt bad because they didn’t have the ‘means’ “they gave me the bigger slice, it was not me” for violating their diet. It seems we are trying to lose the accountability of violating the diet by trying to find external reasons. You don’t see this effect on unrestricted eaters, they are well able to regulate themselves and their caloric intake. It is worth thinking though, maybe those who are restricted may already have a problem with appetite. I wonder if we could have more granular data on this, for how long or how frequent you are restricted. I would also like to know the results of their bloodwork…

The next is a paper on the effect of appraisal-tendency theory on endowment effect Heart Strings and Purse StringsCarryover Effects of Emotions on Economic Decisions . Appraisal-tendency theory claims that specific emotions possess distinct cognitive dimensions (e.g., certainty, control) that conditions us to appraise future. Emotions that occurred a moment ago ‘carry over’ to unrelated subsequent events and affect our judgement. For example anger evokes appraisals of certainty and individual control, thus evoke more optimistic risk estimates and risk-seeking choices. No wonder why we make a lot mistakes when we are angry… Endowment effect is placing higher value to an owned object than would if not owned. If you own the object, you place higher value to it as if it is different from other objects that are not owned. If it is the first time you buy that object, you would place less value to it.

In the study they induce sadness and disgust with videos in subjects to see how the placed value of objects change. In case of disgust, endowment effect is eliminated. In other words, if you experienced disgust just before buying or selling an object, you no longer place the object you own a higher value, i.e. it reduces the prices when selling. It lowers the price for buying as well (compared to control group). It is explained as disgust induces goal to expel. The proximity to the object (owning is closer in that sense) increases the ‘contamination’ with emotions thus the desire to expel increases. Thus it evens selling and buying prices out.

Sadness on the other hand is different. It reduces the selling prices (you no longer see that you own as more precious) but increase the buying prices (first time you buy the object). It is explained that sadness induces goal to change one’s circumstances. Thus buying an object (change in the circumstances) becomes more valuable! You are more willing to pay more to change your circumstances. In a sense, it reversed the endowment effect. They also point out that compulsive shoppers tend to experience depression and shopping tends to elevate depressive mood. It maybe because their circumstances — even though they buy new things — don’t change… It also is shown that anti-depressants decrease compulsive shopping.

So, emotion with same valence-hedonic tone,characteristic of emotions that determines their emotional affect (intrinsic appeal or repulsion)- can have opposing effects on our decisions.

Stewart Brand readings, on maintenance, social capital

The Soul of Maintaining a New Machine is mainly about anthropologist Julian Orr’s study of Xerox service technicians because it has been observed that core of their operational knowledge was social meaning that the hardest problems were solved with technicians discussing the matter, exchanging their know-how (war-stories).

It is intriguing to see how institutions — mostly through high caliber managers — are resistant to change. It is not a surprise but it may be worse than one may think. It is also daunting to further read about how engineers — who write manuals for repair — are clueless when it comes to the ‘field’, actual problems at customer’s side. The manuals were complex decision trees that do not touch upon why a problem occurs to let the technician learn and reason about internals and behaviors of the machine but simply what happened. The main aim for this seems to have control over employees — through knowledge — to mitigate necessary skill thus capital to hire them, they wanted to de-skill them. Contrarily, the social nature actually demonstrates the fact that the job needs highly skilled technicians and the information transfer was crucial for successful maintenance and repair of printers. When reported with this fact, managers instead though “the study revealed how much time — and company money — the technicians were wasting just chatting with each other instead of doing what they were paid to do”. Unsurprisingly, managers immediately suggested company could save serious money but cutting the “chit-chat” down. While socializing, the technicians develop a collective pool of practical and critical knowledge.

Orr and his colleagues tried to build a radio like linking system — named Eureka project — so that technicians can talk to each other with the intention to amplify sharing and increase the reach of information. This idea later on evolved into a knowledge bank. While presenting the idea, it became visible that managers seem to have an image of service technicians as “lowest levels of the corporate hierarchy”. They didn’t want to invest in them. They underestimated them.

Each win on that regard, backfired with de-skilling attempts from managers or trying to cut costs by laying off some of the technicians. The attitude of managers “the lowly technicians should be guided solely by their superiors, not by each other” stagnated bests ideas with great potential. Maybe middle managers were scared and took this as a threat since fundamentally those developments changed their job description and takes their ‘power’ away.

Orr’s also concluded that the asymmetry between technicians and managers was too strong, they had the money and power.

Despite all the resistance, Orr and his colleagues managed to install and make those systems work — after the attempts with the radio —. They created a system where technicians share their information in forms of advice or past cases, effectively creating social capital. Those who contributed were shown respect. Sadly, the developers of the system noted that the technicians and their lore remained isolated from the rest of the company, no additional process were built to incorporate information in Eureka back into the documentation or to provide feedbacks for engineers, designers — in forms of modifications — or even sales people — in forms of leads to suggest an upgrade to a customer— .

It seems Eureka project became a pillar for forums today. We need systems that enable us to notice and honour people with superior skill, show us ways to how to become one. It was a great read to think about environments with collaborative competition and how it fosters competence and mastery. Today, we don’t see much, competition is stripped off of collaboration. A facade of competition drives capitalism as if it is possible to compete with giants. All of the AI companies now try to solve the same problems behind closed doors by throwing expensive compute stealing our precious resources, wasting them through duplicate effort. I am a strong believer of collaboration. I believe that it is a way to greater society.

Unlike financial forms of capital, social capital is not depleted by use; it is depleted by non-use.

From the book Maintenance: Of Everything by Stewart Brand (I could not read the whole yet), I read The Maintenance Race . One of my friends mentioned the book, so I wanted to get a hold on it immediately. It was an incredible reading. It is heavy in sailing terminology, I had to read it by googling stuff so that I can understand what is going on. Stewart Brand thinks and writes about maintenance heavily. It made me realise a lot of things and most importantly I don’t think of it enough. This piece is about the attempt for a new world record by several sailors in 1968. “To make the first solo voyage by sailboat all the way around the world without stopping”. Contenders had to circumnavigate without physical assistance from outside. They cannot further supplement fuel, food, water, or equipment after their departure, they had to be prepared from the start. There are 3 contenders who drew attention the most. An optimist, a pessimist and a guy with “whatever comes, deal with it.” mentality. From now on, I should warn you, there are heavy spoilers ahead…

The optimist was Donald Crowhurst. He is a brilliant mind, incredibly skilled on electronics. He managed to build a long radio from what he had with him in the middle of the ocean. He tended to underestimate the need for maintenance though, even more resist doing it. Steward Brand emphatised with him, he says, their sense of the world may be offended by the grimm reality that everything constantly is decaying. It made me think that I also have a part in me sharing the same. From Crowhurst’s story, it becomes evident that we must find ways and means to accept this impermanence.

It reminded me of wabi-sabi . I went to a kintsugi workshop a year ago. Kintsugi is a practice to emphasise flaws, the points of breaking. In a way, it accepts them, make peace with dullness of the world. From another angle, in terms of hormesis , it is an euphoric feeling to see that sometimes, with a proper stressors, we become stronger from where we are hurt. It is important though to not get ‘broken’ or at least fully to be able to heal. Interestingly, being human is connecting with each other from where we are broken. Trauma-bonding maybe… We desperately need that where we are broken — when seen — is accepted by who we care about, at least our loved ones. Maybe I needed to polish the acceptance with some romanticism, a way of growing into something ‘good’ with scars.

The skill and mindset for not doing important things half-heartedly became central to me. Crowhurst, in his journal during his circumnavigation attempt, diligently made lists of projects needed to be done. He has done a few of them half-heartedly, and abandoned them when his interest dissipated.

Nevertheless, there is no need to praise or adorn what we have to accept any further. He was under-prepared. Crowhurst hoped for the best. It killed him.

The sailor with “whatever comes, deal with it.” mentality was Robin Knox-Johnston. He had a different relationship with maintenance. He thinks doing maintenance cures depression. Thorough-preparation equipped him so that he can made do and mend, repair and solve every problem the wild ocean threw at him. He never was hunted by lack of crucial materials or tools. He left things until they emerge to be dealt with. He threw himself happily into the problem when he had to cope with it. He preserved his energy, and he could do that because he was prepared thoroughly. Knox-Johnston’s style was: “Whatever comes, deal with it.” And he did. He won the race.

The final contestent is Bernard Moitessier, the pessimist. He prepared as if everything would go wrong. He in a sense dealt with maintenance in advance, keeping the need for maintenance at minimum. If there is maintenance to be made, it had to be easy. His boat is designed in a way that it is highly durable, simple to take care of and be low-maintenance. His obsession about details made him the most successful maintainer. He kept everything minimal, even discarded most of the stuff he brought with him on board — including the motor —, emptying the boat to the absolute basics. Less is more, in this case it also meant less maintenance. He actually was going to break the 2 world records at once, he started later than Knox-Johnston but kept up with him, he was going to have the record for the fastest circumnavigation as well. But he realised the peace he had found. He didn’t want to return to the ‘snake-pit’ Europe. He changed course and prolonged his journey, letting go of the race. It didn’t mean anything to him anymore, he has found the most important things, freedom and peace. Moitessier prepared for the worst. It freed him.

A simple yet elegant picture of life unfolds with this story. Assuming that we are not optimist enough to be killed by it. I think it is not enough to just find peace or freedom. A mind that must be able to take it. Otherwise there is no peace or freedom to be found. Evidently, both peace and freedom are not binary, as everything. You have some degree of them. We seem to deal with what haunts us like Knox-Johnston deals with trouble, by solving problems, it feels good and is distracting. Problems induce sufficient amount of freedom, though you have constraints. Solution brings some degre of peace. So, We need problems. It seems to be finally free, to pass through the threshold, we need to have enough experience under our belts to not need problems further, outgrowing them. Getting accustomed to more freedom each time. The more battles we give, maybe, we also are getting more accustomed to peace in between them. So maybe, times we are tight of freedom and peace reversely introduce more freedom and peace in different forms, by changing the height of the bar. When problems are solved, battles end, we find ourselves in the middle of what we are fleeing of, more freedom and peace. We cannot bear it sometimes, it is too quite, too calm. Then maybe, the war at the moment was never finished in the first place. We were fighting the wrong war. Can the capacity to solve problems, fighting wars also flourish a mind that can finally be free and in peace? Is it possible to — if it can — to realise that in the middle of everything?

It is evident to me that the war is not only within. Some ‘teachings’ claim that it is me who is fighting, no one else is fighting me back. I don’t think the war is only within. I disagree, there are fights outside and not only because we need them.

Note: I think maintenance applies to self. Actually, we are the ones who need maintenance the most. Externalising it is missing the point. The more we have the more we need. Maybe we can first start with deloading as Moitessier did, decluttering so that we need less. Not to the fullest, one step at a time, for parts of our lives where maintenance is needed the most. It saddens me that self-sacrifice with hustle culture and so on, is leading the memetic war. We no longer prioritise self-maintenance, contrarily we are taking pride of our decay. What follows is overworked, flattened and drained minds raising to positions of power and determining our future.

Self-reference

I thought about self-reference as well this week. It is not trivial to conceptualise for me. How do I self-reference exactly? How is it done, not specifically a mind but anywhere? It makes me think of dichotomy between meaning and naming. Meaning has information about what is underlying, naming does not. Venus, Morning Star and Evening Star are all the names for the same celestial body. Although Venus seems to carry more information, because if you know it, immediately you can remember that it is the second planet from the Sun, it is the hottest planet and so on. It is actually because we learn all those meanings and name it Venus, as the first indirection. For someone who didn’t know, Venus has meaning as much as Morning Star’s. So, in that sense self-reference, can have 2 forms in that regard, semantic (meaningful bundle of information) or syntactic (referring to the former).

In programs for example, functions recurse through calling themselves via their symbols. It is just a naming. It is a reference and thus it is not wholistic. The function does not know what it is calling actually. In fact, an encapsulating ‘outside’ is necessary where name resolution is necessary, the program counter (the pointer where the next instruction is) should be moved to the beginning of the function again with new arguments. This seems like a nice example for naming. Can the information of what the name points to be imbued in what is referred? Can a table mapping namings to meanings, have a name that has itself as an entry? For example, if the meaning is inscribed as ‘this table’, would it do it? What does ‘this’ point to then? In terms of meaning? Does the table also have to include the meaning of ‘this’? Then also the ‘table’? But then, what the table actually consists of (what it is mapping, how is it mapping) is lost. For true meaning of the self, how inclusive the meaning entry should be? This is maybe getting into the 2 Dogmas of empiricism land.

Meaning seems to be different. It is a collection of names of which we know the definition of, which are also more collections of names. For example, there are ways of referencing in programming with self idiom. The compiler ‘knows’ of a construct referring to itself. It can use that reference to access internals of itself. We can argue it still does not fully exclude naming. The syntax enables semantics, self.component means the ‘component’ within the self. The operator . — which itself is another naming — connects those names with meaning. But is it given or performed by the compiler. I wonder if we can totally remove naming when we think of self-reference in the sense that when I refer to myself, I truly capture whatever I am wholistically, without aliasing. It seems to me that we cannot. We may need a different kind of language for that maybe. To me it is cognitive, there is no way of around it.

Another interesting concept is a quine ^ℹ a program without taking any parameters, produces a copy of itself. It does not have to be identical, the Quine Relay project is a cyclic quine where each time it is run, it spits out a program with identical semantics as itself but in a different programming language. The cycle consists of 128 languages in total. It is an interesting structure. It obviously does not have a clue of what the underlying semantics are. It has indirect self-reference — after the cycle it produces itself — and a direct one semantically at the same time. We can argue how much it can preserve the semantics while cycling but a trivial answer would be ‘all’ since it can generate itself again. Does a quine — direct or indirect — refer to itself?

Another dimension is within the qualia, within our experience there seems to be something that we call I. We have a model in our heads for our positions and kinematics for example. Is this self-reference? It seems so, does moving my arm have to have an element of self-reference? In terms of biology, it seems blurry. As far as our motor systems are concerned, our limb is just a different target that a signal must be passed out to. It does not require self-reference. It only seems so within our experience, our consciousness has spatio-temporal elements.

In addition, more abstractly we realise our emotions. They are experienced as if they appear suddenly, sensations in our bodies, then a mood that has more elements than simple sensations. A voice that seems to narrate all those. Besides ‘reduced’ — I mean as in gastronomy, reducing a sauce — instances consisting of feelings, physical sensations and thoughts, I cannot find self-reference. I am not even sure if I am looing at the right place, why would any biological entity has self-referential elements? Consciousness ,in terms of compressing information about us, seems like it should have self-referential elements but they all are vague ideas. To me, the concept of I as self-reference can only be found in consciousness.

There are lots of reading material about this. The question I am pondering is — maybe a trivial one — whether it is possible to self-reference without external knowledge resolving the reference, as in programming or mathematics, $f(x) = f(x-1) + f(x-2)$ needs external name resolution, within $f$, there is no knowledge of what $f$ refers to or names. Analogously, as dot product needs an additional orthogonal dimension, the product is the magnitude of the vector orthogonal to both vectors, self-reference needs such additional ‘external’ dimension for resolution. It has to lose some information of itself.

I will be reading and thinking about this. I have lots that do not sit right.

My weekly AI smear campaign

📌 I wrote about AI before On Universal Approximation Theorem, more like an initial step of smear campaign for LLM… It also is an incredibly poorly written post, but whatever.

I have seen the post We just found out our AI has been making up analytics data for 3 months and I’m gonna throw up. Specifically, a comment that articulated what I could not:

If there is one single sentence I need everyone to know about LLMs it is from an MIT paper on them… LLMs are correct only as a coincidence, never by intention. I’m paraphrasing but that’s basically it. They will never intentionally be correct or incorrect, it is just a coincidence if they are right. The only output they are programmed for is confidence and agreement.