iiinova — where your digital self keeps evolving

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 09/10/2026

Conveniently dropped right as I'm working on my lecture on this!

140

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 08/10/2026

I've finally lost enough weight for my cold tolerance to start being affected and, well, I'm cold

021

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 04/10/2026

I haven't received many AI-generated emails from students, but I'm totally behind this. I think AI tools are amazing, but probably the most important thing I screen for in these emails is genuine intentionality, and having AI generate the message just totally takes that away.

030

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 01/10/2026

Ever spend an hour reading about pinnipeds on Wikipedia instead of working on your lecture slides

091

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 29/09/2026

On a follow-up note, did y'all know that the chloroplasts of brown algae have four membranes and have an endosymbiont within an endosymbiont? Like you gotta laugh at this point

040

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 28/09/2026

working on my lecture on the origins of eukaryotes and I feel like we all in evo bio kinda take endosymbiotic theory for granted, like that shit is crazy

1101

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 28/09/2026

Iirc in the latest Anthropic report more than half of the work to improve Claude is now being done by Claude itself

2140

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 28/09/2026

Of course, in TNG, they are also in the early stages of having AI systems that are sentient and granted legal personhood in Data and Lore. I think we are far off from this still, but I don't think it's impossible in principle! And they do a great job exploring those implications as well.

200

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 28/09/2026

It's not a replacement for human expertise, either. The ships are manned with trained scientists, engineers, and humanitarians, even though the corpus of knowledge contained in the ship's computer vastly outweighs them. The computer is used like a tool to augment their skills.

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 28/09/2026

They understand the psychological impacts of becoming too dependent on generative AI technology. There are multiple episodes that explore this issue specifically!

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 28/09/2026

These computers can also generate small amounts of matter in literally any possible configuration out of energy, and can do sandbox simulations that are literally physically indistinguishable from reality. Despite this, very little abuse of these systems is depicted.

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 28/09/2026

They are capable enough that they could run a starship entirely autonomously and follow human directives from some remote command center. Despite this, people in the Federation risk life and limb to serve as crew on these ships. They understand that the human experience has to come first.

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 28/09/2026

As is so often the case, I feel like Star Trek, and particularly The Next Generation, presents the best model for how humanity can coexist with AI systems. The Federation ship computers are basically more powerful versions of the LLMs we have today (give me a little rope here).

181

Reposted by Mark Hibbins, Great Lakes Swashbuckler

Joe Hanly @hanliconius.bsky.social · 22/09/2026

I hate to do this with real data, but this is a persistent data vis issue in chromosome synteny riparian plots that we Need to get better at. Left, a synteny plot published today. Right, a rearranged version with exactly the same gross arrangement but with reordered chromosomes:

chromosome synteny riparian plot in which every chromosome has crossing lines

chromosome synteny plot in which no chromosomes have crossing lines. The underlying structure is identical.

29639

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 23/09/2026

I personally have seen multiple philosophers on this website in the past few days say that even making this proposition is nonsensical

010

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

Having clear and precise definitions is important to any discussion of complex and important topics

200

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

Ah so you were also being condescending to a bunch of unengaged viewers of our conversation, fair enough

000

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

I think we can agree that immigration is a high stakes issue for society without taking Trump's stance on the issue. And I'm actually aware of what Mad Libs are, but thanks for the bonus condescension I guess

120

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

Sorry, are you arguing that AI is a low stakes issue because the stakes of individual disagreements about the topic are low? By that logic literally nothing is a high stakes issue for society at large

191

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

Understanding the capabilities of the tools that are completely overtaking our economy and society at an extraordinarily rapid pace is not a "low stakes issue"

271

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

I honestly thought my response was pretty tame and critical of a particular idea, not even close to an attack on your person. But I guess we can agree to disagree

000

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

It's not condescending to suggest that I know literally nothing about tools I've taken courses on and used in published research because I didn't know about a single concept in a related subfield?

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

You actually don't even really need to know what a cell is to grasp the basics of evolution as Darwin described it, so there are a lot of ways that conversation could play out that would make total sense

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

Well, I'm not up on my machine theory but I understand AI prediction models pretty well and have used them in my research. Calling LLMs "some linear algebra" is pretty ridiculous.

110

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

If someone without my background was trying to understand what a cell is so they could have a discussion about what kinds of systems could be considered alive or evolving, I don't think I'd find that very unreasonable

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

I agree that's dumb, but it's also true that AI makes that task way easier for the vast majority of people who just want a company website and don't necessarily need or want all the skills that go along with doing it themselves

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

Well lots of people are going to continue using it as a personal assistant so we should add some security checks

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

I'm totally in favor of AI regulations, I think letting it run rampant on all our systems with no security checks is dumb, but I also think banning it outright is also kinda silly

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

There are a lot of use cases where the tradeoff makes sense, I understand where you're coming from truly but a lot of people really do just need code that's good enough and works

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

I agree with this as well, there is good research suggesting it's bad for educational outcomes

110

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

This is exactly the misalignment problem that AI people are worried about though, the AI hasn't been told that it shouldn't delete your files so it sees it as a reasonable action to achieve its objectives

200

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

That's a bit dishonest, a lot of the stuff the agents did in Hugging Face was not part of the explicit instructions, more of a misalignment issue

000

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

Well sure, but they weren't framed around the actual tangible possibility of it happening like they are now

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

So you disagree with Bernie Sanders calling for an AI slowdown after Hugging Face and related incidents? Just want to understand your perspective, a lot of sensible people seem to be worried about security risks

200

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

I agree with this as well if we're talking about truly general machine intelligence that can do stuff in the physical world

010

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

That's a fair point, I do still think the pace has accelerated quite a bit in the last decade or so

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

Well we've gone from having no LLMs at all to having serious debates about whether machines can be conscious in less than a decade so I'd say the pace is picking up

210

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

Well, 4 billion years of natural selection versus 100 years of computer science is hardly a fair fight lol

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

I don't think either of those things make it impossible for computers to reason, we don't even understand enough about how consciousness and reasoning arise in the human mind to say that it's only possible in biological systems

241

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

Individual neurons don't understand anything either, they are simply reacting to electrochemical inputs. One could argue that "understanding" is an emergent property of complex systems

120

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

The Chinese room experiment is a thought experiment in philosophy, there are no results from the perspective of computer science that say it's actually impossible for a computer to reason

131

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

I don't have access to the full article at the moment but I can check later!

020

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

Idk if "blind" is the right word to describe this particular study design, but still

120

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

Frontier models are literally outperforming medical doctors on diagnosis, there are blind studies on this

2112

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

I guess I still don't know what it means for something to have a "state". If you mean like, a moment-to-moment configuration that changes in response to inputs, then sure I agree. But I don't see how that could preclude self-awareness of some kind if it emerges from the weights

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

I also want to question the claim that self-awareness is a property that's literally impossible to ascribe to an LLM. I do not think they are stateless. My stance is not that I believe they are self aware

100

Mark Hibbins, Great Lakes Swashbuckler @rejectresubmit.bsky.social · 22/09/2026

I think the justification is that LLMs are pretty clearly capable of doing a lot of things that only human brains are otherwise capable of doing, regardless of how reductionist your description of them is

100