weekend ai reads for 2024-08-02

Nearly half (47%) of employees using AI say they have no idea how to achieve the productivity gains their employers expect, and 77% say these tools have actually decreased their productivity and added to their workload.

Microsoft offers to predict how long it will take to complete particular work based on past data on work activities and “AI” models. It outlines possible reasons for deviations between predicted and previously specified durations by suggesting, for example, that a particular client, region, weekday, task or worker will likely increase the time required to carry out the work. As such, it may accuse workers of being slower than expected. This functionality can help dispatchers “enhance their team’s performance”, according to Microsoft.

Humanity has degraded to the point where people are finding better options digitally.

There is a lot of new stuff to build and I think even if the progress on the foundation models kind of stopped now — which I don’t think it will — I think we’d have like five years of product Innovation for the industry to basically figure out how to most effective use all the stuff that’s gotten built so far. But I actually just think the foundation models and the progress on the fundamental research is accelerating so that it’s a pretty wild time.

I’ll argue that, while I am uncertain about whether AI precision farming technology will be good or bad for animal welfare on a per-animal basis, I think we can be much more confident that it will be bad on a total population basis, since it will likely increase the total number of farmed animals. Largely in view of this argument, but also for political feasibility reasons, I think we should consider advocating for restrictions on precision livestock farming in industrial settings.

Wiens says iFixit has since added the crawl-delay extension to its robots.txt. “Based on our logs, they did stop after we added it to the robots.txt,” Wiens says.

Our grade: D

Really helpful in isolated work, but can’t handle a full project or overcome complex tasks.

However, we additionally find that when access is subsequently taken away, students actually perform worse than those who never had access (17% reduction for GPT Base). That is, access to GPT-4 can harm educational outcomes. These negative learning effects are largely mitigated by the safeguards included in GPT Tutor. Our results suggest that students attempt to use GPT-4 as a “crutch” during practice problem sessions, and when successful, perform worse on their own. Thus, to maintain long-term productivity, we must be cautious when deploying generative AI to ensure humans continue to learn critical skills.

They’re thinking about prompt engineering and ChatGPT or Gemini. They’re thinking about, which school tools do I use? But the literacy that people need to have is a broader, more foundational understanding of AI and how does it work as tools or systems that use data to make predictions. What are the things that I need to be aware of or considering when I evaluate the tools that I might want to use?

… nearly 2 in 3 employers said candidates “should have foundational knowledge” of generative AI tools. More than half of employers said they were more likely to interview and hire applicants with AI experience.

This post focuses on the overall architecture for deploying AI applications. It discusses what components are needed and considerations when building these components. It’s not about how to build AI applications and, therefore, does NOT discuss model evaluation, application evaluation, prompt engineering, finetuning, data annotation guidelines, or chunking strategies for RAGs.

Gretel Navigator’s synthetic data generation outperformed OpenAI’s GPT-4 by 25.6%, surpassed Llama3-70b by 48.1%, and exceeded human expert-curated data by 73.6%.

Based on this observation, we propose Self-Route, a simple yet effective method that routes queries to RAG or LC based on model self-reflection. Self-Route significantly reduces the computation cost while maintaining a comparable performance to LC.

Sweetspot leverages AI to help you find, manage, and bid on federal, state, local, and education government opportunities.

A few years ago, I read an article from Janette Fuccella explaining a Priority Framework used at Pendo. It’s based on two main criteria to evaluate the best way to plan your work: problem knowledge and risk. The rest is all about critical thinking.