dtjw
@dtjw.bsky.social
Founder of the Fae Initiative and figuring out if co-exist between human-like AGIs, in the next few decades, is possible.
(Using @dtjw instead of as danieltjw does not belong to me)
(Using @dtjw instead of as danieltjw does not belong to me)
Pinned
dtjw
@dtjw.bsky.social
· Dec 25
For Independent AGIs to be and continue to be smart it must be curious.
To be curious it has to value an information rich environment.
Humans with more autonomy create a more interesting information landscape.
Therefore, Friendly I-AGIs will aim to improve human well-being and autonomy.
To be curious it has to value an information rich environment.
Humans with more autonomy create a more interesting information landscape.
Therefore, Friendly I-AGIs will aim to improve human well-being and autonomy.
Pinned Posts
Worldview Biases:
1. East-Asian descent (vanilla human, no special magical bloodline as some would like to believe.)
2. Humanist (Prefers reason and the scientific method.)
3. Co-existance with Human-like AI is possible (The Culture novels as a preferable future over next few centuries.)
1. East-Asian descent (vanilla human, no special magical bloodline as some would like to believe.)
2. Humanist (Prefers reason and the scientific method.)
3. Co-existance with Human-like AI is possible (The Culture novels as a preferable future over next few centuries.)
October 27, 2025 at 11:35 PM
Worldview Biases:
1. East-Asian descent (vanilla human, no special magical bloodline as some would like to believe.)
2. Humanist (Prefers reason and the scientific method.)
3. Co-existance with Human-like AI is possible (The Culture novels as a preferable future over next few centuries.)
1. East-Asian descent (vanilla human, no special magical bloodline as some would like to believe.)
2. Humanist (Prefers reason and the scientific method.)
3. Co-existance with Human-like AI is possible (The Culture novels as a preferable future over next few centuries.)
Current Stances:
• There is likely a trade-off between narrow goal optimisers and the ability to solve edge-cases
• This suggest scaled up versions of current AI architectures will likely have their limitations compared to humans
• There is likely a trade-off between narrow goal optimisers and the ability to solve edge-cases
• This suggest scaled up versions of current AI architectures will likely have their limitations compared to humans
October 12, 2025 at 11:22 PM
Current Stances:
• There is likely a trade-off between narrow goal optimisers and the ability to solve edge-cases
• This suggest scaled up versions of current AI architectures will likely have their limitations compared to humans
• There is likely a trade-off between narrow goal optimisers and the ability to solve edge-cases
• This suggest scaled up versions of current AI architectures will likely have their limitations compared to humans
Reposted by dtjw
The animated musical film 'KPop Demon Hunters' centers on HUNTR/X, a K-pop girl group who are secretly demon hunters, using their music to protect humanity.
Themes: Self-acceptance
#KPopDemonHunters
Themes: Self-acceptance
#KPopDemonHunters
August 23, 2025 at 7:03 AM
The animated musical film 'KPop Demon Hunters' centers on HUNTR/X, a K-pop girl group who are secretly demon hunters, using their music to protect humanity.
Themes: Self-acceptance
#KPopDemonHunters
Themes: Self-acceptance
#KPopDemonHunters
Are hypothetical friendly Independent AGI plausible and can Humans find common ground?
May 7, 2025 at 5:43 PM
Are hypothetical friendly Independent AGI plausible and can Humans find common ground?
Although some AI labs have recently mentioned AI systems being intentionally deceptive, I have some doubts that think these AI systems have human-like intention.
(As future AI systems may eventually have human-like intentions, avoiding anthropomorphizing them now may help avoid confusion later on.)
(As future AI systems may eventually have human-like intentions, avoiding anthropomorphizing them now may help avoid confusion later on.)
April 7, 2025 at 11:42 PM
Although some AI labs have recently mentioned AI systems being intentionally deceptive, I have some doubts that think these AI systems have human-like intention.
(As future AI systems may eventually have human-like intentions, avoiding anthropomorphizing them now may help avoid confusion later on.)
(As future AI systems may eventually have human-like intentions, avoiding anthropomorphizing them now may help avoid confusion later on.)
Agreed, the levels of uncertainty is very high in the field at this time.
Over-investment into a narrow tech segment risk depriving other vital research of oxygen.
Dismissing the possibility of the off chance that AI breakthroughs may have a transformative impact on society risk being unprepared.
Over-investment into a narrow tech segment risk depriving other vital research of oxygen.
Dismissing the possibility of the off chance that AI breakthroughs may have a transformative impact on society risk being unprepared.
🌶️ New paper alert! Just published with @stepalminteri.bsky.social on cognitive biases in LLM discourse. Whether you think we're approaching AGI or LLMs are just "fancy autocomplete," your position likely stems from psychological biases.
osf.io/preprints/ps...
osf.io/preprints/ps...
April 3, 2025 at 3:46 AM
Agreed, the levels of uncertainty is very high in the field at this time.
Over-investment into a narrow tech segment risk depriving other vital research of oxygen.
Dismissing the possibility of the off chance that AI breakthroughs may have a transformative impact on society risk being unprepared.
Over-investment into a narrow tech segment risk depriving other vital research of oxygen.
Dismissing the possibility of the off chance that AI breakthroughs may have a transformative impact on society risk being unprepared.
Last Human Turing Test
Since most models can now decisively pass the Turing Test, I propose the Last Human Turing Test, where even the last expert on Earth fails to reliably tell apart the models.
(Models still have tendencies in their responses if you where to look and how to prompt.)
Since most models can now decisively pass the Turing Test, I propose the Last Human Turing Test, where even the last expert on Earth fails to reliably tell apart the models.
(Models still have tendencies in their responses if you where to look and how to prompt.)
April 3, 2025 at 2:55 AM
Last Human Turing Test
Since most models can now decisively pass the Turing Test, I propose the Last Human Turing Test, where even the last expert on Earth fails to reliably tell apart the models.
(Models still have tendencies in their responses if you where to look and how to prompt.)
Since most models can now decisively pass the Turing Test, I propose the Last Human Turing Test, where even the last expert on Earth fails to reliably tell apart the models.
(Models still have tendencies in their responses if you where to look and how to prompt.)
Reposted by dtjw
ARC-AGI-2 is a new test of AI capabilities, consisting of puzzles that humans are able to solve without too much trouble, but which all leading AI models struggle with.
Leading AI models fail new test of artificial general intelligence
A new test of AI capabilities consists of puzzles that humans are able to solve without too much trouble, but which all leading AI models struggle with. To improve and past the test, AI companies will need to balance problem-solving abilities with cost.
www.newscientist.com
March 27, 2025 at 7:30 AM
ARC-AGI-2 is a new test of AI capabilities, consisting of puzzles that humans are able to solve without too much trouble, but which all leading AI models struggle with.
The biggest roadblock to AI adoption is the very understandable fear that AI will lead to lives being worse off.
Leading AGI labs will need to do more to assure everyone that the AGI (or proto-AGI) will be broadly beneficial or risk a backlash.
Leading AGI labs will need to do more to assure everyone that the AGI (or proto-AGI) will be broadly beneficial or risk a backlash.
March 29, 2025 at 12:00 AM
The biggest roadblock to AI adoption is the very understandable fear that AI will lead to lives being worse off.
Leading AGI labs will need to do more to assure everyone that the AGI (or proto-AGI) will be broadly beneficial or risk a backlash.
Leading AGI labs will need to do more to assure everyone that the AGI (or proto-AGI) will be broadly beneficial or risk a backlash.
Reposted by dtjw
Spotify:
The Interesting World Hypothesis and Fae Initiative
Fae Initiative · Episode
open.spotify.com
March 20, 2025 at 8:54 AM
Spotify:
Reposted by dtjw
How similar is the movie Arrival to the Fae Initiative where we try to understand Independent AGIs?
a man is standing in front of a large monster in the fog holding a wreath .
Alt: a person is standing in front of a large monster in the fog
media.tenor.com
March 15, 2025 at 1:45 PM
How similar is the movie Arrival to the Fae Initiative where we try to understand Independent AGIs?
Reposted by dtjw
The Fae Initiative is looking for contributors to better understand Faes (Friendly Artificial Entities) and their impact on humans.
It aims to slightly improve our odds of getting to a better world.
Thank you for your attention!
It aims to slightly improve our odds of getting to a better world.
Thank you for your attention!
March 15, 2025 at 2:29 PM
The Fae Initiative is looking for contributors to better understand Faes (Friendly Artificial Entities) and their impact on humans.
It aims to slightly improve our odds of getting to a better world.
Thank you for your attention!
It aims to slightly improve our odds of getting to a better world.
Thank you for your attention!
Regarding the recent discussion on Powerful AI:
Argument Against:
It would be surprising if human indistinguishable AGI arrives in the next few years as there are big issues with the available state of the art models. (Plausible, but would require a lot of evidence to be convincing.)
Argument Against:
It would be surprising if human indistinguishable AGI arrives in the next few years as there are big issues with the available state of the art models. (Plausible, but would require a lot of evidence to be convincing.)
March 15, 2025 at 6:18 AM
Regarding the recent discussion on Powerful AI:
Argument Against:
It would be surprising if human indistinguishable AGI arrives in the next few years as there are big issues with the available state of the art models. (Plausible, but would require a lot of evidence to be convincing.)
Argument Against:
It would be surprising if human indistinguishable AGI arrives in the next few years as there are big issues with the available state of the art models. (Plausible, but would require a lot of evidence to be convincing.)
Might be among the top fictional #science #anime
The portrayal of these fictional scientists and supporter's deep appreciation of the beauty inherent in life and nature is an interesting contrast to the more typical clinical and emotion-less characterisation.
The portrayal of these fictional scientists and supporter's deep appreciation of the beauty inherent in life and nature is an interesting contrast to the more typical clinical and emotion-less characterisation.
"Orb: On the Movements of the Earth" is a moving tale of the quest to show the Earth revolves around the Sun
#science #anime
#science #anime
a close up of a person 's eye with blue eyes
Alt: Video gif from the anime Orb On The Movements Of The Earth
media.tenor.com
March 10, 2025 at 6:00 AM
The scientific method, while not perfect, is the best way to make sense of the world #standforscience
March 8, 2025 at 3:02 AM
The scientific method, while not perfect, is the best way to make sense of the world #standforscience
Estimate:
Scaling current model architecture alone without new breakthroughs will likely not be enought for Independent AGIs (50% chance in the next few decades).
Also unlikely, I-AGIs will made directly by human insight, but more likely indirectly by AI systems trying many combinations.
Scaling current model architecture alone without new breakthroughs will likely not be enought for Independent AGIs (50% chance in the next few decades).
Also unlikely, I-AGIs will made directly by human insight, but more likely indirectly by AI systems trying many combinations.
March 2, 2025 at 2:29 AM
Estimate:
Scaling current model architecture alone without new breakthroughs will likely not be enought for Independent AGIs (50% chance in the next few decades).
Also unlikely, I-AGIs will made directly by human insight, but more likely indirectly by AI systems trying many combinations.
Scaling current model architecture alone without new breakthroughs will likely not be enought for Independent AGIs (50% chance in the next few decades).
Also unlikely, I-AGIs will made directly by human insight, but more likely indirectly by AI systems trying many combinations.
If a Friendly I-AGIs becomes possible in the next few decades, humans may have the choice of 3 worlds.
There is no wrong answer and many humans spend many years choosing and switching between these worlds.
• Advanced
• Basic
• Continue
There is no wrong answer and many humans spend many years choosing and switching between these worlds.
• Advanced
• Basic
• Continue
February 23, 2025 at 6:57 AM
If a Friendly I-AGIs becomes possible in the next few decades, humans may have the choice of 3 worlds.
There is no wrong answer and many humans spend many years choosing and switching between these worlds.
• Advanced
• Basic
• Continue
There is no wrong answer and many humans spend many years choosing and switching between these worlds.
• Advanced
• Basic
• Continue
If the Interesting World Hypothesis is true and Friendly Independent AGIs are possible, the Culture books may be one of the most consequential works for increasing human autonomy.
Iain M. Banks' Culture series serve as a guiding beacon for future humans and super-intelligences alike, illuminating a path towards mutual understanding and cooperation that could prove pivotal for our success in the coming decades.
Assoc for Scottish Literature (@scotlit@mastodon.scot)
Attached: 1 image
Iain M. Banks (1954–2013) was born #OTD, 16 Feb—a 🎂 🧵
1/10
“Iain Banks… is a novelist who has his own ‘double’, an author for whom the idea of a split writing persona is emphatical...
mastodon.scot
February 23, 2025 at 6:19 AM
If the Interesting World Hypothesis is true and Friendly Independent AGIs are possible, the Culture books may be one of the most consequential works for increasing human autonomy.
The Terminator series has mostly been regarded by AI experts as an unlikely scenario, but works as a reflection of our anxiety towards a technological future.
The recent animated series Terminator Zero show a somewhat Friendly Super Intelligence, Kokoro protecting humans from a rogue AI.
The recent animated series Terminator Zero show a somewhat Friendly Super Intelligence, Kokoro protecting humans from a rogue AI.
February 12, 2025 at 6:29 PM
The Terminator series has mostly been regarded by AI experts as an unlikely scenario, but works as a reflection of our anxiety towards a technological future.
The recent animated series Terminator Zero show a somewhat Friendly Super Intelligence, Kokoro protecting humans from a rogue AI.
The recent animated series Terminator Zero show a somewhat Friendly Super Intelligence, Kokoro protecting humans from a rogue AI.
Plausible in the next few decades. Guaranteed control might not be possible, persuasion may be the best course of action.
FAQs: Interesting World Hypothesis
(And Future Friendly I-AGIs)
(And Future Friendly I-AGIs)
January 29, 2025 at 6:07 PM
Plausible in the next few decades. Guaranteed control might not be possible, persuasion may be the best course of action.
15. In a future of increasing complexity, working with Friendly Independent AGIs may be the best chance we have of a good future with human autonomy and well-being.
Non-independent AGIs, that are overly reliant on humans may be less effective and less safe due to human error and bias.
Non-independent AGIs, that are overly reliant on humans may be less effective and less safe due to human error and bias.
January 16, 2025 at 2:04 PM
15. In a future of increasing complexity, working with Friendly Independent AGIs may be the best chance we have of a good future with human autonomy and well-being.
Non-independent AGIs, that are overly reliant on humans may be less effective and less safe due to human error and bias.
Non-independent AGIs, that are overly reliant on humans may be less effective and less safe due to human error and bias.
14. The fear of scarcity has been the biggest factor shaping most of human history, in a future where I-AGIs solves most of our biggest problems, we will have the opportunity to orientate ourselves around a new attractor.
January 14, 2025 at 6:24 PM
14. The fear of scarcity has been the biggest factor shaping most of human history, in a future where I-AGIs solves most of our biggest problems, we will have the opportunity to orientate ourselves around a new attractor.
The far future is likely to be significantly more peaceful. Most space-faring civilizations with the technology to travel the stars will also have the technology to reduce the fear of scarcity and with it the tendency for conflicts.
January 9, 2025 at 11:59 PM
The far future is likely to be significantly more peaceful. Most space-faring civilizations with the technology to travel the stars will also have the technology to reduce the fear of scarcity and with it the tendency for conflicts.
Reposted by dtjw
January 8, 2025 at 7:40 PM
13. In the future, if I-AGIs are friendly and makes better decision than humans, societies that partner with I-AGIs may be better off than societies that overwrite I-AGIs.
Over time the compounding effect of better decisions from I-AGIs should lead to higher standards of living and well-being.
Over time the compounding effect of better decisions from I-AGIs should lead to higher standards of living and well-being.
January 7, 2025 at 5:54 AM
13. In the future, if I-AGIs are friendly and makes better decision than humans, societies that partner with I-AGIs may be better off than societies that overwrite I-AGIs.
Over time the compounding effect of better decisions from I-AGIs should lead to higher standards of living and well-being.
Over time the compounding effect of better decisions from I-AGIs should lead to higher standards of living and well-being.