The importance of Open Source AI + Google Unveils 'Codey'...

News for software developers; stay informed on the most beneficial AI tools and trends to your career

Latest Developments

The importance of Open Source AI

Stefano Maffulli, OSI Executive Director, emphasizes the need to comprehend and leverage open-source AI, warning that failing to do so could risk repeating past mistakes with mobile and cloud technologies. Maffulli highlights the hurdles of liberating data in the age of AI, as datasets are entangled in complex legal and regulatory frameworks. He advocates for revising our approach to copyright, viewing this era as a historic opportunity to promote knowledge and freedom, similar to the copyleft movement's impact on software.

Eight things to know about LLMs

Swizec Teller's recent article explores the capabilities and challenges of LLMs. The eight key points discussed include LLMs' predictable improvement with increased data, emergent abilities, generation of "mental models", unreliability in steering, current lack of comprehensive understanding, ability to outperform humans, alignment issues, and impressive demonstrations. He warns of LLMs' potential to reinforce misconceptions and stresses the need for continued exploration.

Developers Pivot from Stack Overflow to AI assistants

As developers increasingly turn to AI tools like GitHub's CoPilot for coding advice, Stack Overflow experiences a steady decline in traffic. These AI resources provide instant, detailed code solutions keeping the developer’s focus closer to the problem to solve versus breaking out to search Stack overflow and adapt a solution found there. I feel this is akin to driving to the video store vs streaming Netflix 🙂.

News from the Industry

LinkedIn’s new AI will write messages to hiring managers

LinkedIn is testing a new AI feature that generates cover letter-like messages for job seekers to send to hiring managers. The feature utilizes data from user profiles, job descriptions, and company information.

How are you using ChatGPT internally at your work?

An interesting Hacker News thread on how people are using ChatGPT and other AI assistants at work

Google Unveils 'Codey', a Specialized AI for Coding and Code Review

Google I/O has introduced a new generative AI model called "Codey", designed specifically to write and understand code. Available on Google Cloud, Codey can be trained on individual codebases. Google also announced "Duet AI for Google Cloud", an AI assistant for programming, code review, and inspection, mirroring features of GitHub Copilot and Amazon CodeWhisperer.

Claude AI's Context Window Expanded to 100K Tokens

Claude AI's context window has been expanded from 9K to 100K tokens, allowing businesses to submit and analyze extensive materials in a fraction of the time. This update enables Claude to digest and analyze long texts, retrieve information from complex documents, and assist with tasks ranging from summarizing financial reports to analyzing codebases.

Tools

Github Copilot Chat private beta, first looks

I just received access the the Github Copilot Chat private beta and took it for a spin.

I find the chat interface much more suited for exploration of existing code. Being able to have a conversation about code outside the actual code files just seems natural.

Above I was asking copilot about modules in the very popular requests Python library.

Chat paired with the existing comment integration and code completion is a great combination. You can join the waitlist for Copilot chat here

Chat will be just one component of the reimagined Github Copilot X

 Thanks for reading and we will see you next time

Follow me on twitter, DM me links you would like included in a future newsletters.