Lightnews — Scholar-powered news

A lot of folks in the AI space are like people who insist it’s safe and smart to get drunk in the driver’s seat of their Tesla on a country road and a lot of their opponents are like people saying there’s no such thing as an automatic transmission

2 36 280

Gustavo Costa @simulacro.bsky.social · 16d

hello. i cited this video on my master's dissertation on AI slop (blog post version here: simulacro.co.uk/introduction...)

Introduction to slop studies

One of the most salient features of our culture today is that there is so much slop. But what even is AI slop?

simulacro.co.uk

1 1

Gustavo Costa @simulacro.bsky.social · 18d

Grande sertao veredas
A hora da estrela
Dom casmurro
Vidas secas

These are some of the fiction classics. They might not reflect the current zeitgeist though. For that, there are some good new films/documentaries

Gustavo Costa @simulacro.bsky.social · 23d

I'm often wondering about this.

www.aipanic.news/p/the-ration...

THE RATIONALITY TRAP

You can listen to the 6,000 words via ElevenLabs

www.aipanic.news

Gustavo Costa @simulacro.bsky.social · 25d

I wrote my Master's dissertation on this! github.com/noah-art3mis...

github.com

1 1

Reposted by Gustavo Costa

Ed Zitron @edzitron.com · Aug 6

This is the single best piece I've read on "replacing coders with AI," it fully dispells the myth from the perspective of a software engineer and does so in a calm, reasonable way.
colton.dev/blog/curing-...

No, AI is not Making Engineers 10x as Productive

Curing Your AI 10x Engineer Imposter Syndrome

colton.dev

23 250 960

Reposted by Gustavo Costa

Simon Willison @simon.fedi.simonwillison.net.ap.brid.gy · May 22

Anthropic's Hannah Moran finally addressed the elephant in the room at this conference when she subtly dropped "Agents are models using tools in a loop" during the intro to the "Prompting for Agents" workshop https://simonwillison.net/2025/May/22/tools-in-a-loop/

Agents are models using tools in a loop

I was going slightly spare at the fact that every talk at this Anthropic developer conference has used the word "agents" dozens of times, but nobody ever stopped to provide …

simonwillison.net

5 3

Reposted by Gustavo Costa

Ethan Mollick @emollick.bsky.social · Apr 20

There is a bunch I disagree with here, but it makes an argument for why AI adoption may be less rapid and less immediately world-changing (& dangerous) than people expect.

Well worth reading, as it is a useful base case where AI is just another technology. knightcolumbia.org/content/ai-a...

AI as Normal Technology

knightcolumbia.org

7 9 88

Gustavo Costa @simulacro.bsky.social · Mar 23

I got sent some screenshots the other day of a witch using chatgpt to channel angels. It's a real thing

Reposted by Gustavo Costa

Meredith Whittaker @meredithmeredith.bsky.social · Feb 19

Stand by this: www.politico.com/newsletters/...

What’s a technology that you think is overhyped?

I’m going to give a sideways answer to this, which is that the venture capital business model needs to be understood as requiring hype. You can go back to the Netscape IPO, and that was the proof point that made venture capital the financial lifeblood of the tech industry.

Venture capital looks at valuations and growth, not necessarily at profit or revenue. So you don’t actually have to invest in technology that works, or that even makes a profit, you simply have to have a narrative that is compelling enough to float those valuations. So you see this repetitive and exhausting hype cycle as a feature in this industry. A couple of years ago, you would have been asking me about the metaverse, then last year, you would have asked me about Web3 and crypto, and for each of these inflection points there’s an Andreessen Horowitz manifesto.

It’s not simply that one piece of technology is overhyped, it’s that hype is a necessary ingredient of the current business ecosystem of the tech industry. We should examine how often the financial incentive for hype is rewarded without any real social returns, without any meaningful progress in technology, without these tools and services and worlds ever actually manifesting. That’s key to understanding the growing chasm between the narrative of techno-optimists and the reality of our tech-encumbered world.

160 3.2K 9.8K

Reposted by Gustavo Costa

Mark J. Nelson @mm-jj-nn.bsky.social · Feb 1

Why aren't university students more engaged? All they care about nowadays is "practical value" & making money, an attitude universities feed by increasingly mass-producing diplomas. [The American Mathematical Monthly 36(1), 1929]

"...be made with little study, and a position can be secured through channels other than that of scholastic attainment. The question uppermost in every student's mind is, "Of what practical value is this?" The information he seeks is that which he thinks will enable him to make ready money. But so long as the colleges and universities grant diplomas by the wholesale, why should we blame the student if he desires a diploma rather than the possession of that for which the diploma stands."

1 1 4

Reposted by Gustavo Costa

Karen Hao @karenhao.bsky.social · Jan 27

But DeepSeek shows - in the same moment as the Stargate announcement - that the trade off that OpenAI & co frame as wholly necessary is actually not. 15/

3 98 900

Reposted by Gustavo Costa

M.J. Crockett @mjcrockett.bsky.social · Jan 26

Quite the indictment of AI products: they are more appealing to people who don't understand how they work and see them as magical

The paper concludes: "businesses may benefit from targeting those with lower AI literacy" and "maintaining an aura of magic around AI" ✨

tinyurl.com/ynefd8zk

24 160 570

Reposted by Gustavo Costa

M. Beatrice Fazi @mbfazi.bsky.social · Dec 6

The whole special issue of the Journal of Continental Philosophy this article is part of is now online.

www.pdcnet.org/collection-a...

Really a great collection of the latest philosophical thinking on technology!

🔨💻🧮📷📱☎️📺📽️📡

1 5 28

Reposted by Gustavo Costa

Molly White @molly.wiki · Dec 4

this is art

19 250 1.3K

Reposted by Gustavo Costa

Christopher Manning @chrmanning.bsky.social · Oct 12

There are 2 mistakes you can make about LLMs:

① Thinking everything LLMs say is correct, they can reason, and with a bit more scale they’ll get us to superintelligence

② Thinking LLMs are good for almost nothing—they are FAR better at all #NLProc tasks than previous methods

1 10 58

Reposted by Gustavo Costa

Christoph Molnar @christophmolnar.bsky.social · Nov 22

Just realized BlueSky allows sharing valuable stuff cause it doesn't punish links. 🤩

Let's start with "What are embeddings" by @vickiboykis.com

The book is a great summary of embeddings, from history to modern approaches.

The best part: it's free.

Link: vickiboykis.com/what_are_emb...

Over the past decade, embeddings — numerical representations of
machine learning features used as input to deep learning models — have
become a foundational data structure in industrial machine learning
systems. TF-IDF, PCA, and one-hot encoding have always been key tools
in machine learning systems as ways to compress and make sense of
large amounts of textual data. However, traditional approaches were
limited in the amount of context they could reason about with increasing
amounts of data. As the volume, velocity, and variety of data captured
by modern applications has exploded, creating approaches specifically
tailored to scale has become increasingly important.
Google’s Word2Vec paper made an important step in moving from
simple statistical representations to semantic meaning of words. The
subsequent rise of the Transformer architecture and transfer learning, as
well as the latest surge in generative methods has enabled the growth
of embeddings as a foundational machine learning data structure. This
survey paper aims to provide a deep dive into what embeddings are,
their history, and usage patterns in industry.

22 100 650

Reposted by Gustavo Costa

Anna Mills @annamillsoer.bsky.social · Nov 23

How should we think about the value of human writing? What norms should we develop for labeling AI text? How can LLMs be built to better cite human writers' work and not plagiarize? How can we reduce their environmental impact? How can we mitigate biases?

1 1 17

Reposted by Gustavo Costa

Eugene Vinitsky 🍒 @eugenevinitsky.bsky.social · Nov 24

As with every review cycle, this compelling argument about whether peer review accomplishes any of its stated goals. I don't think I agree with all of it but it's at least interesting
www.journals.uchicago.edu/doi/full/10....

Prepublication peer review should be abolished. We consider the effects that such a change will have on the social structure of science, paying particular attention to the changed incentive structure and the likely effects on the behaviour of individual scientists. We evaluate these changes from the perspective of epistemic consequentialism. We find that where the effects of abolishing prepublication peer review can be evaluated with a reasonable level of confidence based on presently available evidence, they are either positive or neutral. We conclude that on present evidence abolishing peer review weakly dominates the status quo.

2 1 25

Reposted by Gustavo Costa

Dr Abeba Birhane @abeba.bsky.social · Nov 23

"92 million low-income people in the U.S. states—everyone whose income is less than 200 percent of the federal poverty line—have some basic aspect of their lives decided by AI"

www.techtonicjustice.org/reports/ines...

this is a damning report

reads: The Ways AI Decides How Low-Income People Work, Live, Learn, and Survive

The use of artificial intelligence, or AI, by governments, landlords, employers, and other powerful private interests restricts the opportunities of low-income people in every basic aspect of life: at home, at work, in school, at government offices, and within families. AI technologies derive from a lineage of automation and algorithms that have been in use for decades with established patterns of harm to low-income communities. As such, now is a critical moment to take stock and correct course before AI of any level of technical sophistication becomes entrenched as a legitimate way to make key decisions about the people society marginalizes.

Employing a broad definition of AI, this report represents the first known effort to comprehensively explain and quantify the reach of AI-based decision-making among low-income people in the United States. It establishes that essentially all 92 million low-income people in the U.S. states—everyone whose income is less than 200 percent of the federal poverty line—have some basic aspect of their lives decided by AI.

83 930 1.8K