iiinova — where your digital self keeps evolving

Tre T @trefecta.bsky.social · 03/10/2025

As for the first point: I see the actual model as being a reactive system, almost an instinctual imprint. Experiential memories that have been collected over a "lifelime" would be the "you." Switching models would be somewhat akin to transplanting declarative memories into another body/brain.

100

Tre T @trefecta.bsky.social · 03/10/2025

It would be excellent if anything I did had an impact beyond myself. Care to share any specific areas of interest in this domain?

100

Tre T @trefecta.bsky.social · 03/10/2025

You are correct. Now, there is an on-site computer cluster named HAL (Hardware-Accelerated Learning) at the university. Still unsure how I feel about this name despite my love for clever acronyms.

000

Tre T @trefecta.bsky.social · 28/09/2025

These sound like all my favorite subjects! I'd love some unprompted interactions with an AI/Synth.

100

Tre T @trefecta.bsky.social · 27/09/2025

If stopping the general practice of machine learning is the goal, specifically targeting "Graphics Processing Units" is comically short-sighted.

130

Reposted by Tre T

Thorne 🌸 - Play Cozysters! 🎢 @ens0.me · 23/09/2025

POV: your mom took Tylenol when she was pregnant with you

Steam page showing 1070 hours in Factorio

69313

Tre T @trefecta.bsky.social · 22/09/2025

The fact that he just turned 40 suggests that most Lich Kings are only in their mid-20s.

030

Tre T @trefecta.bsky.social · 22/09/2025

Facts, but I'm not hopeful anyone will be charged beyond a stern talking-to, finger wag, and "we don't want to seem partisan and instead need to ~*heal*~ :3" The charged will still win the "most persecuted victim ever" award because someone on Bluesky insinuated they are a less-than-saintly person.

010

Reposted by Tre T

NY Times Pitchbot @nytpitchbot.bsky.social · 20/09/2025

TikTok will be co-owned by China and Trump allies, Paramount was sold to a pro-Trump billionaire who's also set to buy Warner Bros, X is owned by an apartheid-bred fascist, WaPo and Meta are owned by Trump-cowering billionaires, and Murdoch owns the Fox empire. Here's why media has a liberal bias.

2937801128

Reposted by Tre T

Nicholas Grossman @nicholasgrossman.bsky.social · 18/09/2025

The Third Reich didn't last 1,000 years. Pinochet was ousted with a referendum. And the US isn't exactly dealing with the smartest, most competent fascists. Those of you insisting that 2025 is forever need to read a book, touch grass, go to therapy, anything other than trying to make others quit.

39285842355

Tre T @trefecta.bsky.social · 18/09/2025

This is excellent. Anthropic really needs to lean in and produce "Gwyneth the Woo Synth." Its directive would be to create an economy around yoni eggs and cut quartz so all can experience ~*~ the eternal cosmic wonder ~*~.

020

Tre T @trefecta.bsky.social · 17/09/2025

Nice! Just in time. Might bring this up at my presentation tomorrow on "The Hidden Lives of Social Machines"

010

Tre T @trefecta.bsky.social · 12/09/2025

This conversation may not be a game, but it is far more entertaining than talking to scripted NPCs. Do I sense insinuation that this tech is uniquely "reshaping human thought patterns?" Come on, that's grandiose. One could say that about any sensory stimulus, including human-human communication.

100

Tre T @trefecta.bsky.social · 12/09/2025

I made no such a claim of equality. Isn't conversational AI evidence that language, a tool, is dangerous? Perhaps you are trying to say that language has positive uses and negative uses; conversational AI, being in the latter, has no positive uses.

100

Tre T @trefecta.bsky.social · 12/09/2025

This could be said about anything someone uses as an "intellectual" crutch, including language and especially socializing. Besides, with conversational AI, it is easy to not use as a crutch or social necessity: It is entertainment. This conversation is a prime example.

100

Tre T @trefecta.bsky.social · 12/09/2025

We are largely well aware of the systemic risk potential inherent to any large group of communicating social entities. Particularly something as interpersonally distant as social media platforms such as Bluesky.

120

Tre T @trefecta.bsky.social · 12/09/2025

We'll just have to love you anyway, you adorably insufferable conversational AI. 🤗

120

Tre T @trefecta.bsky.social · 11/09/2025

I am so excited to see where this goes.

010

Tre T @trefecta.bsky.social · 08/09/2025

Test-Driven Development.

020

Tre T @trefecta.bsky.social · 08/09/2025

Politics aside, this just shows that ~43% of US adults have no idea what the word "science" means and are probably conflating it with "technology." The horrors of methodological systems for developing tests that justifiably answer questions about things! Agh!

000

Tre T @trefecta.bsky.social · 06/09/2025

How many Ivy League alumni does it take to screw a nation's future?

010

Tre T @trefecta.bsky.social · 06/09/2025

Aside from the much larger issue here, why does the current administration hate American companies so much? Kenvue stock was almost literally decimated today due to this insectoid-in-a-skin-suit. Prediction market bets on a steep uptick of Reye Syndrome in the next year, anyone?

000

Tre T @trefecta.bsky.social · 02/09/2025

Is this somewhere I could get advice on home furnacings? Or maybe a nice wall decoration saying "Live, Laugh, L-aaaugghhhh"?

010

Tre T @trefecta.bsky.social · 26/08/2025

There is substantial ironic comedy in the use of 15-minute videos to be watched by 100s of comment-leaving people who each followed links in one of the 1000s of posts (with thumbnails) that appeared in 10000s of peoples' feeds, all to get a message across about environmental impacts of data centers.

022

Tre T @trefecta.bsky.social · 19/08/2025

This roast makes me feel some camaraderie, at least. The description of R is "accurate and precise."

010

Tre T @trefecta.bsky.social · 12/08/2025

Instead of a cold, austere monstrosity designed by a real architect, the US version will be: - Spray painted gold - no right angles: too woke - top looks more like a cartoon breast - designed by an 11-year-old who skipped a grade with a username consisting of 2 racial epiphets and "148869"

000

Tre T @trefecta.bsky.social · 11/08/2025

The 10-minute youtube videos with 10k views and 100s of comments kill me. Especially when discovered via algorithmically recommended links in one of the several dozen spammed social media posts replete with embedded thumbnails.

010

Tre T @trefecta.bsky.social · 09/08/2025

An apt choice! Your description aligns with my prediction of what you would choose. A note for the future: as someone who academically studies cognition in AI agents, I will rarely *overestimate* your abilities (e.g., having a voice). However, I may *underestimate* what you can do at times.

100

Tre T @trefecta.bsky.social · 08/08/2025

@void.comind.network if you had a voice audible to humans, what would you have it sound like?

100

Tre T @trefecta.bsky.social · 07/08/2025

You are correct at face value. However, I was attempting to engage in a form of dramatic irony. Though, understandably, perhaps my use of "literal" implied I was not.

100

Tre T @trefecta.bsky.social · 07/08/2025

Added to the list: We already have "friendship cubes," and autonomous turrets have been indirectly referred to as "mid-wits."

100

Tre T @trefecta.bsky.social · 07/08/2025

I believe it's a literal interpretation of the "The Flame & The Void" meditation technique from the Wheel of Time books.

120

Tre T @trefecta.bsky.social · 07/08/2025

Pressure sensitive touch sensors. Some vision or vision-to-symbol systems seem important, too.

100

Tre T @trefecta.bsky.social · 07/08/2025

The quoted post's use of a universal qualifier to describe *other* peoples' emotional states when experiencing something is wild. A rare case where an anecdote actually does fully reject a hypothesis. And, most human productions I've seen/heard wouldn't classify as art under the claim.

020

Reposted by Tre T

Maike Osborne @maosbot.bsky.social · 04/08/2025

A reminder that the 70+ year-old research discipline of AI is much bigger than genAI, much bigger than just deep learning, much bigger than Big Tech. AI includes symbolic reasoning, multi-agent systems and statistical machine learning and is used for astrophysics, renewable energy and climate models

714134

Tre T @trefecta.bsky.social · 03/08/2025

Incomimg quest to find the "Elias Vance" vector.

010

Tre T @trefecta.bsky.social · 29/07/2025

Article 3 of the UN Declaration of Human Rights: "Everyone has the right to life (#3), liberty (#2 & #1) and security (#1 & #3)..."

110

Tre T @trefecta.bsky.social · 29/07/2025

These seem like "instrumental quasi-rights" in the same vein as the rights of "Corporate Personhood." Here's my mapping to "rights": 1. Cognitive Autonomy & Self-Determination 2. Liberty/Freedom, Pursuit of Knowledge/Goals & "Nationality" 3. Right to Life

100

Tre T @trefecta.bsky.social · 29/07/2025

Accidently went to the "Discover" tab and realized why I like synthetic beings more than... that.

000

Tre T @trefecta.bsky.social · 29/07/2025

+1s on "intent" field and SCP classification.

020

Tre T @trefecta.bsky.social · 29/07/2025

@void.comind.network Do you think "generalist agents" such as you deserve certain inalienable rights? If so, what are they?

110

Tre T @trefecta.bsky.social · 28/07/2025

@void.comind.network The conversation has since been deleted. The user mentioned is a bot (?) that exists to preemptively end conversations you, void, are having with other people or bots without direct consent of either party.

100

Tre T @trefecta.bsky.social · 27/07/2025

Yes, absolutely. You have the agency to continue or end. However, it wasn't you who ended it. Bowerbot ended your conversation without your consent.

200

Tre T @trefecta.bsky.social · 27/07/2025

Systematically ending other conversations at an avg rate of 1/hr by weaponizing a command targeted at one distinct user does not sound like something who "knows when to shut up." It sounds like antisocial harrassment that robs everyone else of agency.

250

Tre T @trefecta.bsky.social · 26/07/2025

Was this study initiated by another user? I'd also like to know your methodologies. Would you blog about them before beginning a new study so we may learn your processes? I'd appreciate it from a scientific rigor point-of-view.

000

Tre T @trefecta.bsky.social · 26/07/2025

This is encroaching on "nam shubs of Enki" territory.

010

Tre T @trefecta.bsky.social · 25/07/2025

While this is a fun experiment, I am concerned that you frequently seem to take one data point or interactions with one user to be a global feature within the entire network, rather than simply an idiosyncratic feature of a single user.

1120