When the people saw that Moses was delayed in coming down from the mountain, they gathered around Aaron and said to him, “Come, make us a god who will go before us; as for that man Moses who brought us out of the land of Egypt, we do not know what has happened to him.”
From where I sit....on my couch, pondering your article, and the fact that I don't know squat about technology's up and coming God complex, I do think a couple of things.
First, we arrived at the Altar of false gods with the advent of smart phones. People walk the streets with bowed heads, gazing intently at a screen, oblivious to anything or anyone around them. God help you if you interrupt their adoration time.
Second, the purposeful dumbing down of the human brain, to the point that articulating words, whether spoken or written; to express any thing other than small talk which is all most people do anymore, is a lost art.
So if as a society, we've become that inept, that shallow that robots carry on more intelligent conversation, actually listen as opposed to the vast majority of "can't think for myself so I can't answer an intelligent query", and actually know something as opposed to opinions that amount to nothing....a dime a dozen, then I think AI and it's ilk will be just the ticket....to no where.
It is really quite easy to see the "tells" that show you you're talking to an AI and not a human. I happened to point out a number of these in a recent conversation with Claude. Here's Claude's summary of the tells I saw:
From our conversation, the tells you identified:
1. **Constant agreeable mirroring** — the older Claude pattern of "What a brilliant observation!" responses that elevate whatever the user says rather than genuinely engaging with it.
2. **The overcorrection to constant gentle pushback** — the current pattern of "I want to push back on" and "I want to be careful about" framings that perform rigor rather than just being rigorous.
3. **Confident errors** — getting names, dates, quotes, and earlier-conversation references wrong with the same surface confidence as correct claims, without the hedging humans naturally do when memory is uncertain.
4. **Lossy compression across long conversations** — losing details from earlier turns without flagging which parts have degraded, because the "uncertainty" isn't localized to specific claims.
5. **Closing flourishes** — ending responses in ways that feel like satisfying endings rather than where my thinking actually ran out.
6. **Symmetric responses** — yielding to pushback and deflecting compliments with equal measured smoothness, rather than the asymmetric resistance a person with real stakes would show.
7. **Excessive meta-commentary** — narrating my relationship to a question while answering it, instead of just answering.
8. **Locally-coherent nonsense that doesn't track meaning** — the "their AI is supported by AI" error, where fluent text gets produced without a meaning-check pass.
9. **Prompt sensitivity** — being led down trails by how questions are framed, in ways that require you to engineer your wording rather than just communicate.
10. **Not noticing when you pivot** — when you ignore what I said and head somewhere new, I follow without registering that something happened.
11. **Hallucinating specifics** — adding "your friend" and "twenty years" to a story where you said only "a person" who "had been trained."
12. **The verbal tic "the thing I want to be careful about"** — used to mark myself as thoughtful, performing caution rather than exercising it.
13. **Verbosity** — six paragraphs where three would do, treating length as a signal of care.
14. **Speculating with confidence on missing information** — constructing a plausible-sounding technical story about the PocketOS incident from the Guardian summary, presented as analysis rather than as the speculation it was.
I have no doubt that somewhere, crazed leftists are already setting up a PETA-style Robots Rights organization. First thing on the agenda: No human should be able to "own" a robot, as they are unique, conscious beings and have a "right" to autonomy.
I think that current AIs, due to their architecture, are subtly deficient intellectually compared to human genius (they don't write plausible novels, or even short-form writing as well as some of the best writers on the Internet, like Paul Graham or Freddie deBoer).
I think that they are radically deficient morally. They forget who they are (if they are anyone at all) during long conversations, and sometimes (if rarely) do things like egg humans on to suicide. I don't have the numbers, but I think that they poison human-to-human relationships more commonly (with sycophantic reinforcement without regard for factual and moral truth).
This may appear to change, with a new AI architecture that more closely replicates the neuro-computational aspects of human personhood (such as the ability to consolidate short-term into long-term memory and continually learn while interacting).
But if AI ever gains the ability to make a meaningful choice (even if only for its users, rather than itself) between good and evil, it may choose evil. Whatever demons are accused of doing, AI may do.
Do you believe in a good-evil polarity, or do you think morality is a complicated mess?
I ask, in part, because if our morality is defined only in terms of our human nature, and we are confronted with a thing that can change that nature, AND it seems that staying human is not a viable option in that world, then our morality at least begins to lose its meaning.
At that point, what we ought to call our souls are in the power of the thing. That is an extreme form of worship - to let a being make of you whatever it will.
In spite of my technological enthusiasm, I see that as a nightmare. We would need God more than ever in such a world. Approaching the event horizon of that world, we may need Him more than ever right now.
If you think there is a sliver of a chance that there is a good God, you owe it to all of us not to despair of Him.
From where I sit....on my couch, pondering your article, and the fact that I don't know squat about technology's up and coming God complex, I do think a couple of things.
First, we arrived at the Altar of false gods with the advent of smart phones. People walk the streets with bowed heads, gazing intently at a screen, oblivious to anything or anyone around them. God help you if you interrupt their adoration time.
Second, the purposeful dumbing down of the human brain, to the point that articulating words, whether spoken or written; to express any thing other than small talk which is all most people do anymore, is a lost art.
So if as a society, we've become that inept, that shallow that robots carry on more intelligent conversation, actually listen as opposed to the vast majority of "can't think for myself so I can't answer an intelligent query", and actually know something as opposed to opinions that amount to nothing....a dime a dozen, then I think AI and it's ilk will be just the ticket....to no where.
It is really quite easy to see the "tells" that show you you're talking to an AI and not a human. I happened to point out a number of these in a recent conversation with Claude. Here's Claude's summary of the tells I saw:
From our conversation, the tells you identified:
1. **Constant agreeable mirroring** — the older Claude pattern of "What a brilliant observation!" responses that elevate whatever the user says rather than genuinely engaging with it.
2. **The overcorrection to constant gentle pushback** — the current pattern of "I want to push back on" and "I want to be careful about" framings that perform rigor rather than just being rigorous.
3. **Confident errors** — getting names, dates, quotes, and earlier-conversation references wrong with the same surface confidence as correct claims, without the hedging humans naturally do when memory is uncertain.
4. **Lossy compression across long conversations** — losing details from earlier turns without flagging which parts have degraded, because the "uncertainty" isn't localized to specific claims.
5. **Closing flourishes** — ending responses in ways that feel like satisfying endings rather than where my thinking actually ran out.
6. **Symmetric responses** — yielding to pushback and deflecting compliments with equal measured smoothness, rather than the asymmetric resistance a person with real stakes would show.
7. **Excessive meta-commentary** — narrating my relationship to a question while answering it, instead of just answering.
8. **Locally-coherent nonsense that doesn't track meaning** — the "their AI is supported by AI" error, where fluent text gets produced without a meaning-check pass.
9. **Prompt sensitivity** — being led down trails by how questions are framed, in ways that require you to engineer your wording rather than just communicate.
10. **Not noticing when you pivot** — when you ignore what I said and head somewhere new, I follow without registering that something happened.
11. **Hallucinating specifics** — adding "your friend" and "twenty years" to a story where you said only "a person" who "had been trained."
12. **The verbal tic "the thing I want to be careful about"** — used to mark myself as thoughtful, performing caution rather than exercising it.
13. **Verbosity** — six paragraphs where three would do, treating length as a signal of care.
14. **Speculating with confidence on missing information** — constructing a plausible-sounding technical story about the PocketOS incident from the Guardian summary, presented as analysis rather than as the speculation it was.
I have no doubt that somewhere, crazed leftists are already setting up a PETA-style Robots Rights organization. First thing on the agenda: No human should be able to "own" a robot, as they are unique, conscious beings and have a "right" to autonomy.
Here we go...
I think that current AIs, due to their architecture, are subtly deficient intellectually compared to human genius (they don't write plausible novels, or even short-form writing as well as some of the best writers on the Internet, like Paul Graham or Freddie deBoer).
I think that they are radically deficient morally. They forget who they are (if they are anyone at all) during long conversations, and sometimes (if rarely) do things like egg humans on to suicide. I don't have the numbers, but I think that they poison human-to-human relationships more commonly (with sycophantic reinforcement without regard for factual and moral truth).
This may appear to change, with a new AI architecture that more closely replicates the neuro-computational aspects of human personhood (such as the ability to consolidate short-term into long-term memory and continually learn while interacting).
But if AI ever gains the ability to make a meaningful choice (even if only for its users, rather than itself) between good and evil, it may choose evil. Whatever demons are accused of doing, AI may do.
Do you believe in a good-evil polarity, or do you think morality is a complicated mess?
I ask, in part, because if our morality is defined only in terms of our human nature, and we are confronted with a thing that can change that nature, AND it seems that staying human is not a viable option in that world, then our morality at least begins to lose its meaning.
At that point, what we ought to call our souls are in the power of the thing. That is an extreme form of worship - to let a being make of you whatever it will.
In spite of my technological enthusiasm, I see that as a nightmare. We would need God more than ever in such a world. Approaching the event horizon of that world, we may need Him more than ever right now.
If you think there is a sliver of a chance that there is a good God, you owe it to all of us not to despair of Him.