Twitxy Saved

Alejandro Piscitelli

@piscitelli

Internetómano. Felinómano. Bibliómano, Tigrenómano @EscenariosUdeSA (2020- ) @UNTREF (2018- 2022) @datosuba (1996-2022) Obsesionado por la Tercera Cultura
Tigre, Argentina filosofitis.com.ar Joined January 2008
498 Following 30K Followers 57K Posts

Latest posts 5

Alejandro Piscitelli @piscitelli
La cibervigilencia es Goliath a la n, las críticas no le hacen ni cosquillas (no ven que no ven). Proliferan libros/slogans sobre cómo debería ser el mundo, cuándo nunca hicieron demasiado para anticiparlo ni menos para prevenirlo, salvo citar a algunos gurúes desconsolados :-(
Mariano Amartino Verified account @amartino Oct 26
Asi como lo ves, este televisor (y cualquier otro Samsung, LG o Vizio) toman screenshots cada 500 milisegundos los procesan y los mandan a un server para que anunciantes y data brokers puedan comprar tu perfil de usuario...

¿que mas hacen los televisores de Samsung? analizan el sonido ambiente, tus charlas, los ruidos y los procesan y los mandan a un server.

¿que mas hacen? los venden en un ecosistema de 400 data brokers (no solo anunciantes ni solo LLMs) ¿que mas hacen? si el microfono no anda buscan uno del sound bar ¿que mas hacen? si enchufas tu Mac via HDMI tambien sacan screenshot de lo proyectos y asi... todos y cada uno de tus datos los venden.

Vizio (oficialmente en su S10 a la SEC) contó que logro 115 millones de USD de ingresos de esta rama de negocios.

Y vos despues te preocupas porque "ChatGPT me lee!"
View quoted post on X
Alejandro Piscitelli @piscitelli
Palabras muy mayores, en vez de empantanarnos en la conciencia (o no) de los LLM's
Pablo Malo Verified account @pitiklinov Oct 3
Victoria de Andrés explica aquí que la inteligencia artificial ya no es solo una herramienta para observar o analizar la vida, sino que está diseñando biología que no existe en la naturaleza. Hasta hace poco, el biólogo encontraba una proteína en la naturaleza, averiguaba para qué servía y, si acaso, la retocaba. Hoy el orden se invierte. Se parte de la función que se quiere y la máquina propone una molécula nueva que la cumpla, sea una proteína, una enzima, un anticuerpo o un fragmento de ARN que no existe en ningún ser vivo:

“Pero es que también la IA está haciendo algo trascendental a este respecto. Tanto, que ya es factible “sustituir” a la propia evolución biológica y hacer en segundos lo que la selección natural ha podido tardar miles o millones de años en conseguir (o, incluso, aún no ha alcanzado). Por ejemplo, para virus tan llamativos como el Ébola o el SARS-CoV-2, se han generado con IA anticuerpos maduros con hasta 7 veces más afinidad por los antígenos virales que los anticuerpos naturales, y con hasta 160 veces más afinidad en el caso de los anticuerpos no maduros. Algo parecido se puede hacer con las moléculas de ARN, cuya evolución se está ya “dirigiendo” en procesos tan importantes para nuestra salud como la activación de procesos cancerígenos.

¿Alucinante? Pues vamos a dar una nueva vuelta de tuerca. El Evo 2 es un modelo de IA entrenado con unos 9 billones de pares de bases de ADN procedentes de bacterias, arqueas, virus, plantas, hongos y animales (incluidos humanos) y al que no se le ha asignado ninguna tarea biológica concreta. Pues bien, ella (o él) solito fue capaz de aprender reglas biológicas directamente de las secuencias de ADN, predecir el efecto de mutaciones y …¡generar ADN nuevo!

Esto no es un mero avance: es un salto brutal desde el momento en que no hablamos predecir biología sino de poder realizar auténtica ingeniería biológica. En otras palabras, hablamos de crear vida.”

theconversation.com/jugamos-a-ser-…
View quoted post on X
Alejandro Piscitelli @piscitelli
Uno de los pocos que hace números y se toma en serio la altísima improbabilidad de que la IA masiva y la democracia (ya bastanta abollada) puedan coexistir. De esto no se se habla, de mil efectos secundarios algo, de utopías fantasiosas, todo el tiempo.
Daron Acemoglu Verified account @DAcemogluMIT Oct 1
Third question on AI.

A question that also remains unasked is whether the AI boom can continue without leading to a massive increase in inequality.

A recent paper by Stijn Van Nieuwerburgh runs the numbers on how much revenue the AI industry needs to generate to recover its massive investment (summary and a link to the paper can be found here: brookings.edu/articles/finan…).

Van Nieuwerburgh’s arithmetic should make us more concerned. AI investments will average about 3.6% of GDP annually between 2025 and 2032. Van Nieuwerburgh calculates that, using a 10% rate of return, the industry would need to generate annual revenues of about $3.7 trillion by 2032 to recover these costs (growing from its current levels of about $200 billion or so). That is significantly more than 10% of current US national income, and will likely remain around 10% of national income by 2032, even if GDP growth rose from its current level.

A large fraction of this revenue will go to capital income. That means a massive increase in the share of capital in national income, which has already risen substantially over the last 25 years or so – now standing at an all-time high of about 47% (bls.gov/news.release/p…).

Capital income is much more unequally distributed than labor income, so a massive increase in the capital share of national income will translate into a very sizable surge in inequality.

The rise in inequality may not stop with the capital share. My work with Pascual Restrepo documents that (automation-driven) increases in the capital share of national income are typically associated with rising labor income inequality as well (see, for example, economics.mit.edu/sites/default/…). The same may happen in the next several years, boosting inequality further.

What is missing from our current debate is any discussion of a fundamental dilemma these numbers pose: can the AI boom avoid both an economically costly crash and a huge increase in inequality?
If the industry reaches these revenues, inequality surges. If the industry does not become profitable, a crash, with substantial costs in terms of lost output and jobs, becomes likely.

My assessment would be that the industry is unlikely to reach levels of revenue Van Nieuwerburgh calculates. First, diffusion has been and will likely continue to be slow. Second, competition from open-weight models, which are getting better, will limit how much proprietary models can charge. Third, despite important advances, I still believe that AI models will not be able to automate entire occupations anytime soon, thus limiting their value to businesses as cost-saving devices.

Whether this leads to a crash or not is more complicated and will depend on whether various AI companies are bailed out and what kind of support they receive.

Nevertheless, even if revenues fall short of these gargantuan amounts and we avoid a dramatic surge in inequality, I expect that the diffusion of AI will push up inequality between capital and labor and within labor.

If inequality does surge, a further question becomes central: can our democracy survive such astronomical levels of inequality?
View quoted post on X
1 replies 7 reposts 34 likes 3.4K views View on X