Matt Halloran

3.5K posts

Matt Halloran banner
Matt Halloran

Matt Halloran

@mdhalloran

Building agent swarm tech | Optimist✨& Futurist👽

Miskatonic University Katılım Eylül 2019
589 Takip Edilen887 Takipçiler
Sabitlenmiş Tweet
Matt Halloran
Matt Halloran@mdhalloran·
I think there’s a surprisingly short path to near-total automation of software engineering. Once agents can iteratively write, test, evaluate, and refactor code inside tight feedback loops, software engineering collapses into an optimization problem. The bottleneck then shifts from execution to planning: mapping the space of possible apps, defining how they interoperate, and distributing work across large fleets of machines. Software becomes cheap, abundant, and mostly invisible. When software can be generated, tested, and modified on demand, the real leverage shifts to understanding how those systems behave. That’s where simulation takes over. Running millions of counterfactual worlds to see how systems interact, fail, reinforce incentives, or spiral in unexpected ways - before they ever touch the real world. By then, the boundary between software, governance, and society starts to dissolve. Most people aren’t ready for how fast that transition could happen.
English
2
0
7
1.2K
Matt Halloran retweetledi
Mookafish
Mookafish@Mookafish·
It's clear that the best way to solve humanity's problems, like climate change and pollution, is to accelerate technological progress so we can find ways to improve and optimise our industries. The idea that we need to slow down or "de-growth" is dumb because it will only limit our access to new tools and resources, actually preventing us from being able to make things better. The Earth could easily support billions of extra humans if we just continue to optimise how we build our homes, grow our food, and generate energy to power our machines. The quoted post is a great example. Pesticides have always been an environmental hazard and a health concern, but the invention of semiconductor materials has allowed us to build light emitting diodes and computers, which lead to these machines which can keep our food clean at scale without introducing unnatural chemicals into the environment.
Danny Bernstein@bernsteind

autonomous robot driving through the field at night. no chemicals. no pesticides. just UV light killing pathogens and pests while everyone sleeps. this is @tricrobotics. this is what chemical-free pest control looks like at scale.

English
15
46
321
9.6K
Matt Halloran
Matt Halloran@mdhalloran·
Do you have any evidence that the boring company ever submitted a bid or had any formal involvement in the California high speed rail? All I see is that his idea of the hyperloop was inspired by his frustration of California’s high speed rail. Any claim that he was involved in it or tried to derail (pun intended) California’s own plans seems like an unfounded conspiracy
English
2
0
19
351
ArmedOctopus
ArmedOctopus@TrueFader·
@mdhalloran @nicochristie He got involved, over promised, and under delivered. The project changed from high speed rail to his proposed hyper loop, using maglev and a vacuum tunnel constructed by his Boring Co. He didn’t have the maglev knowledge or technology and tried to outsource it.
English
4
0
0
668
nico
nico@nicochristie·
Pretty crazy that Elon is only worth $1 Trillion Thats like 200 miles of California high speed rail
English
170
738
14.1K
273.1K
ArmedOctopus
ArmedOctopus@TrueFader·
@nicochristie I know you’re trying to dunk on California, but you know Musk is largely responsible for the delays and runaway costs, right?
English
15
0
8
7.8K
Moon
Moon@Moonicker·
@DoireHendrix @hankgreen but how in the heck can it not feel futuristic!? it's literal full-blown sci-fi, delivered for free to anyone who wants to ... checks notes ... talk to a computer!
English
2
0
1
43
Hank Green
Hank Green@hankgreen·
Today's AI tools don't feel futuristic to me, they feel like using a BBS in 1996. Exciting, but adorably unaware of what they will someday be.
English
57
32
1.5K
69.8K
Matt Halloran retweetledi
Peter Hague
Peter Hague@peterrhague·
The issue of Mars colonisation (or space colonisation in general, if we don't want to get into the planet/habitat debate between space advocates) is surprisingly one of the key dividing lines in present day politics. You can reliably predict someones opinion on it by which party they vote for, by how they feel about Elon Musk as a person, or by their level of optimism/pessimism about technology. There is almost total correlation on believing it is a good/bad idea from a technical perspective and thinking it is desirable/undesirable from a political perspective. Everyone seems to be quite sure that the laws of physics vindicate their opinions on this matter. Its not surprise I'm in the "let's go" faction. But I am increasingly of the view that even beyond the issue of Mars, that faction must win, or else Earth as well will be sunk into the mire of degrowth, envy politics, and misery.
Peter Hague tweet media
English
154
94
1.1K
94.2K
Matt Halloran
Matt Halloran@mdhalloran·
@ID_AA_Carmack I suspect this can be already achieved through a sufficiently advanced harness. Most architectural choices (e.g. screaming architecture, boundaries of responsibilities, testing seams) can already be enforced through a test/feedback loop
English
0
0
0
70
Matt Halloran retweetledi
John Carmack
John Carmack@ID_AA_Carmack·
It seems like LLMs could optimize coding style by exploring ways of structuring code so weaker and weaker models can still successfully perform tasks in a codebase. There are surely stylistic quirks that are peculiarly impactful to transformers, but I bet there would be a lot of overlap with human capabilities. Optimizing for understanding should help even the top frontier models, allowing them to understand things “at a glance” without having to explicitly explore. There will remain “better” and “worse” ways to code.
English
178
110
1.7K
114.8K
Matt Halloran
Matt Halloran@mdhalloran·
If everyone can create civilization-ending externalities from home, society needs an immune system. I think we have to choose between a centralized panopticon vs a decentralized, cryptographically constrained surveillance state. The latter is much harder to set up well, but seems feasible given powerful enough coding agents. It’d require features like decentralized sensors and AI auditors, local-first processing, cryptographic attestations, threshold governance for data reveal/action, and strict auditability over what is retained, queried, or escalated. Basically Flock cameras and sensors, but running provably agreed-upon code and no centralized control
English
0
0
1
42
Henry Dowling
Henry Dowling@henrytdowling·
@tenobrus Completely agree! Idk why there are so many haters to this argument. Like, if everyone in the world has the capability and means to design covid-27 with a 20% CFR, how exactly are we going to survive this? Genuinely asking @ the many people who are disagreeing with this take
English
3
0
15
1.2K
Tenobrus (→vibecamp)
Tenobrus (→vibecamp)@tenobrus·
doing it silently seems bad, but in general i am very happy anthropic is taking strong measures to prevent distillation and progress in open source models, and i hope openai will do the same.
Tenobrus (→vibecamp)@tenobrus

ftr i agree this feels like a pretty bad move. doing it silently doesn't actually seem to add very much safety and adds a bunch of uncertainty and blast radius. i'm on totally onboard with slowing RSI, but just... use a normal refusal??

English
16
6
243
27.9K
Matt Halloran
Matt Halloran@mdhalloran·
@bad_at_schedule @lougrims I took AP physics and wasn’t taught that. I suspect the majority of americans are only taught kinematics, and that’s even if they take physics in the first place. At my school you had the option to take some other science class
English
0
0
2
257
constellator
constellator@bad_at_schedule·
@lougrims What sort of high school physics are you guys learning that doesn't teach stefan boltzmann?
English
19
0
45
44K
Lou
Lou@lougrims·
The whole AI datacenter in space discourse is so bad. You have illiterate AI bros being dunked on by people who took high school physics (space is not cold! vacuum in an insulator!). Themselves being dunked on by STEM graduates (Stefan-boltzmann! only 100m^2 of radiators!). 1/2
English
85
59
3.2K
320.8K
Matt Halloran retweetledi
alphaXiv
alphaXiv@askalphaxiv·
As believers of open research, we are disappointed to see Anthropic silently degrading Fable 5 for AI development "Any topic related to building pretraining pipelines, distributed training infrastructure, or ML accelerator design... may have limited effectiveness through Claude via methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning." Not only do they get to decide what you use LLMs for in research, but this also enables them to silently intervene in your research without you knowing. This sets a dangerous precedent. If a model refuses openly, users can understand the boundary. If a model falls back to another model, users can still evaluate the difference. But if a model silently modifies or weakens its own answers while still pretending to help, researchers lose the ability to know whether a failed result came from their own idea, their implementation, or an invisible intervention by the model provider. That is not safety. Safety policies should be transparent, auditable, and user-visible. On top of that, the people most harmed by this are not the largest labs with massive teams and proprietary infrastructure. It is the independent researchers, academic groups, startups, and open-source builders who rely on public tools to compete, innovate, and pioneer AI for everyone else.
alphaXiv tweet media
English
166
721
3.9K
221.3K
Sean
Sean@sean_from_earth·
The space data center thing is great example of normal people not having read the proper sci-fi cannon. To them it seems needlessly complicated and dumb. To those of us with sufficient sci-fi psychosis, it's perfectly logical and even crucial.
Sawyer Merritt@SawyerMerritt

SpaceX has just officially unveiled its AI1 satellite, the first generation of its AI satellite. Overall Specs: • 150 kW peak compute payload • 120 kW average compute payload • 70 kW per ton • Compute provider interchangeable Dimensions: • Wingspan: 70 meters • Deployed height: 20 meters Thermal System: • 110 m² deployable liquid radiator • Redundant pumping loops • Integrated micrometeoroid shielding • Deployable liquid radiators Solar Power System: • 150 kW solar array • 250 W/m² • SpaceX-manufactured solar technology from Bastrop, Texas Architecture: • Centralized compute module • Large deployable solar arrays • Deployable liquid-radiator thermal management system • AI-focused compute satellite design ("AI1 satellite") Elon: "The AI satellite is much simpler than a Starlink satellite. The AI satellite is essentially a lot of solar cells, you still need some laser links, but you don't have all of the super complex antennas that you have on a Starlink satellite. The easier one to design for is the AI satellite. It's bigger. A lot of this is technology we've already made with the Starlink V3 satellites."

English
35
41
726
38K
Damian Guppy
Damian Guppy@perthguppy·
@sean_from_earth And to those of us with actual experience with datacenters and physics, chucking a rack of gear into orbit that has no other reason to be in orbit is a stupid idea. Each of these satellites can hold a single DGX GB300 rack.
English
4
0
4
459
Matt Halloran retweetledi
Joshua Tindall
Joshua Tindall@jd_tindall·
Recursive self-improvement going from being a ridicule-worthy fringe sci-fi concept to a completely normalized part of the discourse which is “obviously the plan” is one of the more dramatic Overton window shifts I’ve experienced There’s something disorienting about it, like if the sky suddenly turned red, and everyone acted like it had been that way all along
English
69
105
1.3K
73.3K
Luciferous_22
Luciferous_22@Luciferous_22·
@AnechoicMedia_ They aren’t even teaching kids to type anymore so I think we should start there.
English
1
0
3
72
AnechoicMedia
AnechoicMedia@AnechoicMedia_·
For about two decades, "computer literacy" education in America was realized as "learn Microsoft Office". Office ended up having a long reign as a useful thing to know, but this was still a narrow vision of computer proficiency that didn't prepare people well. With AI, I don't know what the narrow skills to teach are. Early AI adopters discovered advanced usage techniques which circulated as tribal knowledge, but these patterns were obsoleted faster than anyone could have turned around a curriculum for the next grade, much less for students to have entered the workforce.
Alec Stapp@AlecStapp

.@tylercowen has the best take I've seen on AI and education:

English
16
8
215
23.5K
Matt Halloran retweetledi
Anthropic
Anthropic@AnthropicAI·
Each time we release a model, we run the same test: give it code that trains a small AI model, ask the new model to speed it up. It takes a skilled human 4-8 hours to reach 4x faster. In May 2024, Claude Opus 4 averaged a ~3x speedup. This April, Mythos Preview achieved ~52x.
English
63
243
3.7K
1.1M
Matt Halloran retweetledi
roon
roon@tszzl·
the frontier labs don’t have “comms problems”. reality right now has a comms problem. what is happening is a little scary and there’s no nice words anyone could say, especially not those profiting from it, that’ll make it feel that much better
English
191
185
3.5K
369.3K
Matt Halloran retweetledi
Neoliberal Akiho ✨🇺🇳🇨🇦
Neoliberal Akiho ✨🇺🇳🇨🇦@NeoliberalAkiho·
It’s beyond me that people can look at spaceflight and not get starry eyed. Must be a miserable experience to not dream of a brighter future for humanity.
English
27
128
1K
12.8K
Quick Thoughts
Quick Thoughts@lthlnkso·
Contra Goldman Sachs CEO I think there is a risk of a jobs apocalypse.
English
23
12
287
11.4K
Matt Halloran
Matt Halloran@mdhalloran·
@Tylerkaerr @jmhorp Maybe most compute is moved to the edge (phones/glasses/robots/iot/etc.), with most improvements coming from online learning instead of pretraining/rl?
English
2
0
2
215
Tyler
Tyler@Tylerkaerr·
@jmhorp Data centers are going to be replaced by... a more efficient and powerful way of processing data... which somehow isn't a data center? I'm open to new ideas here. Maybe I'm thinking like a naive person from 2026. Where do we process data if not in data centers?
English
8
0
71
3.9K
Matt Halloran
Matt Halloran@mdhalloran·
Issue a new currency with a basic income distributed daily. Debt is banned, and existing debt is wiped or causes a capped deduction from income until paid (principal only, no interest). Existing property and wealth are kept for a 1:1 exchange to the new currency. All new investments and capital allocation are handled by AI, with profits distributed equally. All positions of power are replaced with AI to prevent profit generation from prediction markets. With no more sources for debt or reliable profit creation, inflation from minting can balance out all wealth over several generations. Land rights either expire after some long period of time, or landowners simply keep that land forever and other people must build in other areas of the earth, megastructures, or other planets
English
1
0
0
30
Quick Thoughts
Quick Thoughts@lthlnkso·
@skinsnbirds If there was an AI that could do (nearly) all useful economic labor, how should we distribute/allocate resources?
English
3
0
6
487

Keşfet

@nicochristie @TrueFader @Moonicker @DoireHendrix @hankgreen @ID_AA_Carmack @tenobrus @bad_at_schedule