weekend ai reads for 2025-01-10

Because one agent is just software, two agents are an undebuggable mess.

We’re not adding any more software engineers next year because we have increased the productivity this year with Agentforce and with other AI technology that we’re using for engineering teams by more than 30%, to the point where our engineering velocity is incredible. I can’t believe what we’re achieving in engineering.

This means that those with significant capital when labour-replacing AI started have a permanent advantage. They will wield more power than the rich of today—not necessarily over people, to the extent that liberal institutions remain strong, but at least over physical and intellectual achievements. Upstarts will not defeat them, since capital now trivially converts into superhuman labour in any field.

In one particularly stilted exchange about the pumpkin giveaway, Rose asked James, “And how have these free pumpkins impacted the community?” to which James responded, “The free pumpkins have brought joy to many.”

Instead of asking whether machines will ever become conscious, we might ask whether humans can become conscious enough to outgrow the “artificial intelligence” both inside them and in the machines around them.

We test baseline agents powered by both closed API-based and open-weights language models (LMs), and find that with the most competitive agent, 24% of the tasks can be completed autonomously. This paints a nuanced picture on task automation with LM agents -- in a setting simulating a real workplace, a good portion of simpler tasks could be solved autonomously, but more difficult long-horizon tasks are still beyond the reach of current systems.

While we are committed to making these types of investments, we recognize that we cannot do everything ourselves. We need to tap into the platforms and expertise that technology companies can provide, as demonstrated by our recently announced collaboration between Tempus and Medtronic Structural Heart.

What counts as cheating is determined, ultimately, by institutions and examiners. Many universities are already adapting their approach to assessment, penning “AI-positive” policies.

We introduce Interactionalism as a new set of guiding principles and heuristics for the design and architecture of learning now available due to Generative AI (GenAI) platforms. Specifically, we articulate interactional intelligence as a net new skill set that is increasingly important when core cognitive tasks are automatable and augmentable by GenAI functions.

Featuring Garrett Smiley, CEO & Co-founder of Sora Schools, John Danner, Co-founder of Project Read AI and Spark Space, and Ted Dintersmith, Founder of WhatSchoolCouldBe.org

This section will start with an overview of agents and then continue with two aspects that determine the capabilities of an agent: tools and planning. Agents, with their new modes of operations, have new modes of failure. This section will end with a discussion on how to evaluate agents to catch these failures.

A control group of arbitrary phishing emails, which received a click-through rate (recipient pressed a link in the email) of 12%, emails generated by human experts (54% click-through), fully AI-automated emails 54% (click-through), and AI emails utilizing a human-in-the-loop (56% click-through). Thus, the AI-automated attacks performed on par with human experts and 350% better than the control group.

Reclaims the internet by replacing all instances of ‘ChatGPT’ with ‘my stupid friend.’

They used information theory to find patterns in his music that help explain how Bach conveyed messages—including musical, mathematical and emotional information—through his works.