Shipping fast versus writing good code. I think there's a time and place for both. For certain things, my first priority is code quality and craft. For others, my first priority is to get something out in the world as fast as possible. As a general rule, I tend to be very

Avanzamos, con enorme ilusión, en el proyecto de colaboración de Loyola Andalucía con las universidades jesuitas en África. iaju.org/2024/11/11/ajc…

I have now run enough o1 experiments with fellow academics on their hard problems in fields ranging from neurology to genetics to economics that I believe that it has genuine potential to help with science & academics should try it to see. Even when it is off, it spurs thinking.

I really wish there was more interest in benchmarking AI models on real tasks. The fact that there are not 30 different benchmarks from different organizations in medicine, in law, in advice quality, etc. is a real shame, as people are using systems for these things, regardless.

More creative humans generate more creative output from AI image generators, as judged by other people. The effects are highly significant, but relatively modest, suggesting that human creativity matters in getting creative work from AI... but perhaps not as might be expected.

I’ve learned more in 1 month with ChatGPT-4o than years of courses. Here are 8 prompts that can save you thousands of dollars:

El presidente de la Fundación Universidad Loyola, Jaime Oraá SJ, y el rector de la Universidad, Fabio Gómez-Estern, mantienen una audiencia privada con el Papa Francisco Fabio Gómez-Estern uloyola.es/blog/la-univer…

Here’s the thing: studies of social media find that most posts are actually positive, but angry posts spread more widely. Why? Joyful posts only spread within our tight social network but anger & indignation are universal, moving more easily across the weak ties between network

The average IQ of software developers will increase by several points over the next few years. This is just a theory, but I think AI will significantly decrease the number of developers over time. To stay employed, you'll need to be smarter and work harder than ever. I don't

Real world filmed to produce miniature effect using tilt shift lenses

I do not think that AGI-seeking labs are in the best position to leverage AI power. Training NNets for achieving ever more powerful LLMs does not provide better understanding of the open opportunities. There is a gap, that should be addressed by someone else.

Compartí unos días con el padre jesuita Francois Kaboré este año en Burkina Faso. Además de una autoridad académica y social, es un ejemplo de compromiso con su país. religiondigital.org/el_papa_de_la_…

Somewhat scary, to say the least. More so when they blantantly use words suggesting what is mostly uncertain in the AGI future: broadly distributed wealth, prosperity...

Do you know any businesses built by non-technical founders who used AI to build 100% of the system? Around 60% - 70% of the teams I know use AI to augment their people, but I don't know anyone using AI *exclusively* instead of real engineers. If this hasn't happened, I wonder

Thank you, I hope it happens. But free from external interference, pls.

It's funny how some of the most prominent "accelerationists" effectively don't believe in AGI.

Con la IA, sí, programamos varias veces más rápido. Pero también aprendemos varias veces más rápido. Antes nos pasábamos el día en Stack Overflow, MDN, W3C… Pero una parte importante del esfuerzo era localizar qué estudiar y reunir fuentes fiables de las que estudiarlo. Ahora

Ethan Mollick Helpful framing from the new Yoshua Bengio et al report

Hollywood is on the brink of massive change...

AI is unlikely to speed up science by multiples acting solely as a co-pilot to humans. There are some early successes using current LLMs to autonomously research, plan & conduct experiments, including learning how to use equipment by reading documentation, with big implications.