weekend ai reads for 2024-06-21

Focus on giving the agent context and letting it “think”, instead of hoping it gets the answer right in one try.

More interestingly, the knowledge the doctor agents have acquired in Agent Hospital is applicable to real-world medicare benchmarks. After treating around ten thousand patients (real-world doctors may take over two years), the evolved doctor agent achieves a state-of-the-art accuracy of 93.06% on a subset of the MedQA dataset that covers major respiratory diseases. This work paves the way for advancing the applications of LLM-powered agent techniques in medical scenarios.

Build reliable AI agents with monitoring, testing, and replay analytics. No more black boxes and prompt guessing.

The fact that generative AI is advancing and evolving so quickly is why it’s important to be an active participant in defining the future for it.

“People are using AI tools at work as they feel overwhelmed and under duress at work, and they want some relief,” Stallbaumer says, “It’s still early! Companies and people are on the journey…(but) there's a lot of light bulbs going on.”

She notes the recent Microsoft Work Trend Index found three quarters (75%) of respondents are now using AI at work in some way - double the number from six months ago, with the technology helping boost creativity and freeing up time to focus on crucial tasks.

However, conversations with enterprise customers have convinced us that the demand for responsible deployments is real. We believe that Responsible AI is rapidly turning into an area of utmost importance, and we’re excited about investment opportunities in this area.

The popularisation of artificial intelligence (AI) has given rise to imaginaries that invite alienation and mystification. At a time when these technologies seem to be consolidating, it is pertinent to map their connections with human activities and more than human territories.

Scott speaks about ChatGPT Edu, specifically how it will affect higher education and the edtech industry.

While Human-Machine Interface (HMI) traditionally referred to the panels and dashboards engineers use to control a piece of equipment or a factory-floor robot, the HMI we envision here will come in the form of an application software layer with a UI / UX tailored to the use-case.

Any individual data point might be negligible in value to a model’s performance, but collectively, a large group of users can aggregate novel data sets that are valuable for AI training. This is where the idea of data DAOs can fit in. With data DAOs, data contributors could see economic upside from contributing data as well as govern how that data is used and monetized.

You have a fixed cost of ~$100 a month, which results in a ‘profit’ of $57. To make up your initial server cost of $3,800, you’d need about 66 months or 5.5 years to benefit from this approach.

Our position is that open-endedness is a property of any ASI, and that foundation models provide the missing ingredient required for domain-general open-endedness. Further, we believe that there may be only a few remaining steps required to achieve open-endedness with foundation models.

Anyone can create an AI persona, called a Butterfly, in minutes on the app. After that, the Butterfly automatically creates posts on the social network that other AIs and humans can then interact with.