Luis Garicano 🇪🇺🇺🇦 Profile picture
Oct 30 9 tweets 2 min read Read on X
The EU AI Act seems designed to allow AI only for routine tasks while hindering its use in high-level problem-solving.

This will endanger European AI startups and significantly damage EU productivity.

THREAD on our post today in Silicon Continent.
1/9 Image
An AI bank teller in the EU would need two humans to oversee it. A startup building an AI tutor faces countless hurdles before launching. The is the reality under the EU AI Act—a well-meaning but flawed attempt to regulate AI.
2/
siliconcontinent.com/p/the-strange-…
The Act classifies AI systems by risk: unacceptable, high, limited, and minimal. Unacceptable systems, like social scoring or workplace emotion recognition, are banned.
Fines can reach €15 million or 3% of global revenue.
3/
High-risk AI—which includs uses such as education, employment, law enforcement, and more—faces strict regulations.
Before releasing an AI tutor, a startup must build risk management systems, document everything, get certified, and more.
4/
The Act also targets "General Purpose AI Models" like large language models, regulating them by capability, not use. Models trained with over 10²⁵ FLOPs are deemed "systemic risks" and face extra restrictions.
5/
These models must detail their training data, allow copyright holders to opt out, run risk assessments, and monitor their entire lifecycle. If used in high-risk areas, all other heavy regulations apply too.
6/
Enforcement is fragmented. Each of the 27 EU countries will have multiple bodies overseeing compliance, leading to inconsistency and confusion. Staffing these bodies with AI experts is a significant challenge.
7/
Startups will struggle to navigate this maze, giving an edge to big firms that can afford compliance. Innovation may stall, and Europe risks falling behind in AI development and application.
8/
Europe should rethink the AI Act before it fully takes effect, instead of stifling innovation with heavy rules. The stakes are too high to get this wrong.
Read the full post by @pietergaricano.
9/9
siliconcontinent.com/p/the-strange-…

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Luis Garicano 🇪🇺🇺🇦

Luis Garicano 🇪🇺🇺🇦 Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @lugaricano

Oct 16
We keep hearing how solving Europe’s innovation stagnation requires more public spending. But the numbers show otherwise: the EU falls behind in private R&D investment, not public.
A thread based on this week’s blog.
1/10
siliconcontinent.com/p/the-problem-…Image
As a share of GDP, Europe spends 0.74% on public sector R&D, compared to the U.S. 0.69%.

The actual R&D gap is in private sector spending, where Europe spends 1.3% of GDP compared to the United States' 2.4%!
That gap is worth 341 billion in R&D spending in 2021. /2
Take the story of DeepMind. It is a contemporary of EU's AI flagship, the Human Brain Project, launched in 2013 with €600M in public funding, aimed to simulate a human brain in 10 years. It's now widely regarded as a failure, while DeepMind leads in AI. /3
Read 10 tweets
Apr 19
New data shows that the EU Commission has blown the chance the NextGen gave it to get the EU on a growth path. Two key elements.
1. Pensions in Spain.
2. Reforms in italy.

The new data is from the ageing report of the EU Commission on the budgetary impact of the pension "reforms"- more below

( h/t @rdomenechv @fernandosols with official data from the Spanish government.)

Small THREAD (1/7)economy-finance.ec.europa.eu/publications/2…
The EU NextGen plans gave an unprecedented and powerful stick to the EU Commission to demand reforms and investments in exchange of money. Never has the Commission had the chance to get states to get some reforms going.

In Spain, the EU Commission has been complicit (in spite of numerous warnings) in setting Spain on an unsustainable Fiscal path
(2/7)
Under cover (!!!) of the "reforms" required by the European NextGen plan, the Spanish government abrogated the 2012 reforms of pensions (the single reform done by the Rajoy government), based on an automatic adjustment mechanism, without putting anything else meaningful in place.

The cost is 3.3 points of GDP higher than before the reform.

(3)Image
Read 7 tweets
Mar 15
Some reactions to the (wonderful) Levitt interview.
1) On the @uchicago PhD program and the atmosphere in the department in the 90s (toxic?).
2) On Price Theory and its future at @uchicago and beyond.
3) On the "technification" of economics and the blurring of the "theory-empirics" boundaries.
(link to interview: )
(Thread)
1/npodcasts.apple.com/us/podcast/ste…
1) On the Econ PhD Program. I went in 1992, graduated in 1998. I did not feel the ambiance was toxic. It was tough work, almost brutal, not toxic. I was given a chance I would not have gotten elsewhere. There was nothing personal about the standards. We were getting trained by the best and that was intellectually invaluable -we got the chance of a lifetime. Here are some profs of my first two years (note 5 nobels):
Macro: Sargent, Lucas, Cochrane, Woodford, Stokey, Townsend.
Micro: Becker, Rosen, Murphy, Scheinkman
Metrics: Hansen, Heckman, Zellner.
It was extremely hard, by far the hardest thing I have ever done. But it should be hard. They were trying to put a bunch of kids at the frontier of knowledge.
It was not for everyone, but we knew what we were getting into. My admired supervisor, Sherwin Rosen, then department chair, gave us a "superstar" (he wrote THE paper after all) talk on the first day. He told us half of us would fail in the first year Core (and exit with an MA, is that so bad?), half of the rest would not make the prelims. Of the 50 we were there, maybe 10 would finish the PhD, most of those would never get any citation.
And yet we persisted. We wanted to learn, and were grateful for the hance.
2/n
2) On Price Theory. What is the Chicago Price Theory style? Best thing I can recommend is to experience it yourself by listening to the playlist of Kevin Murphy's classes. . He is an amazing teacher, and makes economics come alive.
Is it true as Levitt says, quoting Mulligan to Friedman, that this style of Micro lost in the market place of ideas?
3/nyoutube.com/@chicagopricet…
Read 10 tweets
Oct 1, 2023
Prometí hacer un pequeño hilo con datos sobre el estancamiento de la economía española, al hilo de mi entrevista en @elmundoes.
Aquí van 4 gráficos. Al final del hilo, el texto de la entrevista completo.
La economía crece por demografía o por productividad.
Demografía: somos 47 millones.

La ONU estima que a finales de siglo seremos 30.8 millones, con una orquilla entre 21m y 43m: en el mejor caso (con mucha inmigración) estancamiento de la población.

data.un.org/Data.aspx?q=Sp…
Image
Productividad: estancada desde 2006. Ha crecido 0.5% en total en 16 años. Image
Read 6 tweets
May 22, 2023
We have seen much about Bob Lucas' macro contributions these days, but he also had a highly influential contribution to the theory of the firm: the "assigment theory of the firm", which explains, for instance, why Musk earns so much (and controls so many resources).
THREAD Image
Before Lucas 1978, we had Marshall-Viner: individual firms have U-shaped long-run average cost functions. In equilibrium, each firm produces at the minimum point of this curve, with firm entry or exit adapting to get aggregate production. Resonable for plants, but not for firms! ImageImage
The size distribution which emerges is a solution to the problem: allocate production over firms to minimize total cost.
It goes without saying that this is counterfactual for firm size.
Here is the actual firm size distribution (Axtell, Science 2001). Image
Read 25 tweets
May 4, 2023
🚨Stunning document, leaked today supposedly from GOOGLE, on whether there is a "moat" (Barrier to entry) in the LLM space.
The author argues neither Google nor OpenAI have a moat, and open source wins. Thread with critical comment at end:
semianalysis.com/p/google-we-ha…
The question of whether one-three players dominate the industry (like Operating Systems or Search) or it is perfectly competitive is hugely important:
1) For consumer welfare: more competition is better.
2) For control of direction/ethics: more competition makes it harder.
To answer the question requires answering: what are the barriers to entry protecting the incumbents? What can their competitive advantages be?

If there are huge barriers, then the winner (Google in search) takes it all in AI, like Google did in search or MSFT in Office.
Read 11 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(