In this Q&A episode, the host discusses finding guests algorithmically, exploring impactful technologies and tools, data annotation as remote work, Cue Basic programming language, programming experiences and hacker culture, 'grab' command line utility and the importance of Git for source control.
Read more
AI Summary
AI Chapters
Episode notes
auto_awesome
Podcast summary created with Snipd AI
Quick takeaways
The host of the podcast uses an algorithmic approach to find guests by crawling through the archive at rxiv.org and selecting publications based on specific categories and predicted interest levels.
Data Skeptic is a platform for engaging with diverse researchers and showcasing their work, aiming to create spin-off projects and collaborations in the future.
Deep dives
Finding Guests for the Show
Finding guests for the podcast is primarily done algorithmically. The host crawls the archive at rxiv.org, a preprint platform, and selects publications in specific categories of interest. The text from the publications is extracted and indexed using keywords and BERT embeddings. A machine learning model trained on these embeddings predicts the interest level of each publication. By combining keyword matching and predicted interest, the top results are selected and presented to the host for consideration.
The Journey of Data Skeptic
Data Skeptic is a weekly interview podcast that explores different topics through the lens of contemporary preprint research. Though it started as a hobby project, it has become a platform for engaging with diverse researchers and showcasing their work. The host envisions Data Skeptic as a stage for academics and aims to create spin-off projects and collaborations in the future.
Defining Kyle and His Autobiography
The host, Kyle, prefers to emphasize methodology, technology, and insights rather than personal stories or feelings. However, he lists the top 20 impactful technologies and tools that have shaped his development journey. Ranging from handheld electronic games to Python libraries like Pandas and Matplotlib, these technologies have influenced his approach to problem-solving and coding.
Artificial General Intelligence (AGI)
The host believes that AGI, though not yet achieved, is becoming more tangible. He mentions Pedro Domingos' five tribes of AI, where connectionist approaches seem to hold promise for achieving AGI. With the advancement of large language models, general-purpose learning systems are evolving, demonstrating the ability to learn and reason like humans. The host acknowledges that AGI has not yet passed the Turing test, but speculates that with further advancements and the right combination of approaches, AGI may become a reality.