weekend ai reads for 2024-05-31

The next company who reaches the next major plateau gets to announce a groundbreaking AI, and the second one after that gets to announce something that’s 0.3 percent better. Do you want to be the company delivering groundbreaking AI, or the company, you know, delivering 0.3 percent better?

Advanced computing is now core to the security and prosperity of our nation; we need it to optimize national intelligence, pursue scientific breakthroughs like fusion reactions, accelerate advanced materials discovery, ensure the cybersecurity of our financial markets and critical infrastructure, and more.

A potential negative aspect of understanding that the purpose of a system is what it does, is that we are then burdened with the horrible but hopefully galvanizing knowledge of this reality. For example, when our carceral system causes innocent people to be held in torturous or even deadly conditions because they could not afford bail, we must understand that this is the system working correctly. It is doing the thing it is designed to do.

And fringier, more explicitly ideological outlets on the right have noticed that their businesses — already rocked by an industry-wide decline in web traffic — seem unlikely to get an AI bailout.

It felt like getting a birthday card with only the prewritten message inside, and no added well-wishes from the wisher’s own pen. An item off the shelf, paid for and handed over, transaction complete.

The market for AI in personalized learning is forecast to grow at a CAGR of 44.3% to reach $48.7 billion by 2030 – up from $5.2 billion in 2022.

AI helps 73% of students understand the learning material better and 63% study more efficiently.

“This year, everyone coming in here will have prompt engineering training to get them ready for the AI of the future,” Mary Erdoes, who runs the asset- and wealth-management unit, said at the firm’s investor day Monday.

A narrow singularity is a future point in human affairs where AI has so altered a field or industry that we cannot fully imagine what the world on the other side of that singularity looks like. I think academic research is facing at least four of these narrow singularities.

Unlike other model providers, Meta is not in the business of selling model access via API. So while they'll open-source as long as it is convenient for them, developers are on their own for model improvements thereafter.

That begs the question: if Meta is only pursuing open-source insofar as it benefits themselves, what is the tipping point at which Meta stops open-sourcing their AI? Sooner than you think:

Team-level CI depends on collaboration, understood by behavioral scientists as aligning “mental models” to achieve shared goals. Research shows collaboration is often confounded by our tendency to anchor on information that we already know and our limited ability to fully integrate others’ perspectives into our own mental models, especially when communication load is high.

While LM evaluation is difficult and suffers from a number of challenges as we have described, there are measures that can be taken to significantly improve current practices. We provide our high-level recommendations regarding such measures, and detail our motivations briefly for each.

The figure shows that after a boost in library capital investment, reading test scores steadily increased. In the short run, library investments increased reading scores by 0.01 standard deviations. Seven years out from a project, scores were 0.04 standard deviations higher in districts that invested in public libraries than their counterparts.