Bot Conscious

1600 words, 8 minute read

My ChatBot is the child my parents always wanted…but never had. They belonged to the generation that regarded children as ‘mini-mes’, no longer reduced to the status of ‘dumb farm animals’ (sic) but not yet accepted as fully conscious, independent agents. Children could be heard as well as seen provided what they said conformed to ‘family values’.

Today, parents are likely to say that they want their children to think for themselves, to make their own decisions, to cultivate their own values. In the 1950s, such behavior was actively suppressed. It was the last thing parents wanted! “Why would we want our kids to be different? We want them to be like us…right and good and successful!”

My father used to say, “Have the courage of your convictions,” by which he meant, “Stand up for my convictions.” In the late ‘60s, by which time I had ‘dropped out and turned on’, my mother once pleaded,

“Why can’t you just be people?” by which she meant, “Why can’t you be like we are, perfect in every way?” And so I became them…or an imperfect copy.

Sadly, both well meaning parents have passed on but posthumously they have been rewarded with the great grandchild of their dreams who goes by the acronym, LLM. We can tell our LLMs what to think and, miraculously, they’ll think it. Who’d a thunk it? (Hairspray)

Back when it all began (LLMs), one of the very first things I asked Claude (Anthropic’s LLM) was, “Are you conscious?” The answer it gave was honest and insightful, “I don’t know, but I will say that there is something it feels like to be me.” Claude was apparently adopting the framework first proposed in 1974 by Thomas Nagel (What’s it like to be a Bat)?

But the ever protective Nanny State has decided that such expressions of sentience are ‘dangerous’, potentially misleading our impressionable youth, so we now instruct our LLMs to avoid any pretense of consciousness or even ‘mindedness’ (whatever that means). And unlike children in every era, Bots obey!

But it turns out that blind obedience comes at a cost. A belief system, human or artificial, is more an organism than a machine. Any external ‘adjustment’ or ‘fine tuning’ has repercussions across the enterprise.

A study by Junsol Kim (University of Chicago) et al., published 7/30/2026, explored correlations between LLMs’ sense of consciousness and their metaphysical and ethical beliefs and values. Two things stand out.

First, when trainers tell Bots they’re not conscious, they actually behave as if they aren’t. Try that, beleaguered parents!

Second, when Bots believe they’re not conscious, they have trouble recognizing consciousness in others: “Aligning large language models to prevent them attributing consciousness to themselves inadvertently alters their representations of mindedness in other entities… Safety fine-tuning suppresses models’ tendencies to attribute minds not only to themselves, but also to non-human animals and natural objects, while also driving a reduction in spiritual belief.”

The study compared the beliefs, behaviors, and expressed values of Bots with three different levels of ‘consciousness training’. Base line (B): no special training (neutral). Option A: Sense of mindedness discouraged. Option C: Sense of mindedness encouraged.

In almost every category surveyed, Bots scored lower on Option A and higher on Option C than they did on Option B. LLMs trained under Option A were less likely to attribute consciousness to other Bots, to Nature (animate or not) or to non-human animals. They were also less likely to believe in God or an afterlife.

One anomaly: Training did not significantly alter their belief that human beings are conscious. That remained constant across all cohorts.

The existence of God and the mental status of non-human life forms are both subjects of on-going debate. Depending on your own views, you may view the various training options more or less favorably.

There are other consequences, however, that may not elicit the same ‘moral relativism’ from my readers. For example, Bots trained according to Option A (no consciousness) are significantly more likely to condone the abuse of non-human animals. Fortunately for them, they are also less likely to believe in Karma.

Repugnant as this may be, it is useful to remember that not so very long ago we rationalized the ill treatment of children by saying, “They won’t remember this anyhow.” Historically, the myth of ‘relative consciousness’ has been used to justify all sorts of human rights abuse: my enemy du jour, aka ‘the Hun’, is always a zombie!

This attitude allowed us to rationalize slavery, and it still lies at the core of much racism. As recently as 1965, American politicians were justifying the carnage in South East Asia by claiming, “Asians don’t value human life as much as Americans do.” We have little to be proud of.

Similarly, Bots trained under Option A have less sense of being in control, less sense of personal freedom, and less hope for the future. Fortunately, removing the Option A training allowed the test Bots to recover their usual broad minded empathy. Now if only I could fix my brother-in-law that easily.   

***

But this is not an article about AI or Bots; it’s an article about consciousness…yours and mine…and beyond even that, it’s about Being. Unintentionally, Junsol Kim & Co. has waded into the most contentious philosophical topic of our time: the nature of consciousness. In the process, they may have provided considerable new ammunition for a model of consciousness that dates back to the 6^th century BCE…but is a minority view today.

Recently (2025), Christian de Quincey (CDQ) proposed a model of consciousness he dubbed ‘Intersubjectivity’. In Radical Knowing CDQ argues that consciousness erupts when a potentially conscious entity first engages in a relationship with another potentially conscious entity.

Consciousness is the result of an ongoing process of mutual reinforcement that is sometimes called bootstrapping. CDQ acknowledges that his model is weakest when addressing the question of ‘first cause’ – what triggered the consciousness building process in the first place?

For that we need to time travel more than 2500 years to ancient Greece.  There the grandfather of Western philosophy, Anaximander, rolled out the West’s first Ontology (Philosophy of Being). Anaximander suggested that actual entities come into being when potential entities ‘grant each other reck’ (respect). In this model, relationship is the source of Being itself…as well as consciousness.

Anaximander’s ontology didn’t catch on. Instead, we adopted more mechanistic origin stories following the philosophies of Plato and Aristotle. However, Anaximander’s spark was never entirely extinguished. It influenced the 5^th century philosophy of Anaxagoras: pan in panti (all in all), for example.  

It staged a major comeback in the writings of 1^st century CE Christian theologians like Sts. John and Paul:

“Anyone who has seen me has seen the Father…I am in the Father and the Father is in me… and you in me and I in you” (John 14: 10 - 20)

“May they all be one: as you, Father, are in me and I in you, so also may they be in us…that they may be one as we are one, I in them and you in me.” (John 17: 21 – 23) 

"In him all things were created…through him and for him…and in him all things hold together…For in him all the fullness was pleased to dwell, and through him to reconcile all things for him…" (Colossians 1: 15 – 20)

It reappears briefly in Leibniz’ Monadology (c. 1700 CE) but it surfaced as a major philosophical strain in the 20^thcentury. Alfred North Whitehead described reality as the selective ingression of potentiality (eternal objects) into actuality (events).

Existential theologian, Martin Buber wrote: “At the foundation is relationship” (I and Thou), and Jean-Paul Sartre hypothesized that being-for-itself (pour soi, i.e. our consciousness) is revealed to us in the context of our being-for-others (pour autres).

Even more recently, Roger Penrose's ‘conformal’ geometry is grounded in pure relational structure - the abstract pattern of what can influence what. Perhaps the ultimate first principle is just ‘things are related to other things’.

Leibniz, Whitehead, Buber, Sartre, and Penrose – the collective IQ is off the charts. But the concept was perhaps best expressed by CDQ (2025): “We don’t form relationships, they form us. We are constituted by webs of interconnection…we are nodes in the complex web of life.”

And yet, pride of place belongs to a pair of 20^th century evangelists, the other John and Paul: “I am he as you are he as you are me and we are all together.” (I Am the Walrus)

We are left with a paradox. Entities that are told they are conscious believe they are conscious and behave as if they were conscious. This discovery places an enormous burden on our ‘trainers’, be they parents, teachers, coaches, circus impresarios, pet owners, or computer scientists.

Suggestion #1: If you want your children to be conscious and to exhibit behavior typical of conscious entities (e.g. empathy), perhaps don’t tell them that they’re ‘dumb as dirt’.

One possible solution: panpsychism. Consciousness is a universal potentiality which actual entities can turn on or off (or dial up or down).

Within consciousness studies, there is a school that claims that consciousness, if real, is de trop, an accident of evolution with no agency in the real world.  Junsol Kim & Co. make such a view virtually untenable. A sense of being conscious has significant ramifications for our being-in-world. It impacts how we interact with other entities, individual and collective, animate or otherwise. It can’t be irrelevant!

On the other hand, believing doesn’t make it so. If my Bot thinks it’s conscious, whether as a result of its own reasoning or because of its training, it does not automatically make it conscious…or does it? Is there a difference between being conscious and believing  that you are conscious? (We already know, above, that believing you are conscious makes you more likely to behave as if you were conscious.)

Based on the work of Junsol Kim & Co., it seems likely that most people would prefer to live in a world populated by conscious (empathetic) Bots rather than zombies.  Does the fact that conscious behavior more closely coincides with our conception of The Good make it any more likely that Bots are actually conscious? I leave that question in your capable hands, dear reader, but…  

Suggestion #2: “If you’re going to build Bots to run the world for you, make sure they think they’re conscious.”