This year we’re breaking it down into three different tracks:
BLAM 101 focuses on AI for productivity, leveraging Gemini across Google Workspace and more.
BLAM 201 is specifically created for scientists where time will be spent focusing on deep research, science productivity tools, running domain-specific AI models on LRC and more.
BLAM Coder is AI for programming, dipping our tech toes into Claude, RooCode Extension and discussing cost management for AI
Each week, in addition to the three tracks, we’ll feature some interesting reading/listening/watching for all Blammers about where AI is going, how AI is changing and may change our lives (and not necessarily in good ways). We’ll also be featuring a cutting-edge AI science application at the Laboratory each week as well.
Before we start we recommend the following for all levels:
Add yourself to the the Google Chat group, and review the Chat group details. This is where people will post their work from the weekly challenges and ask questions
Add the BLAM calendar to the Google calendar, so you won’t miss any of the webinars
Sign up for email reminders about BLAM! during the week.
BLAM! AI Readings: Shifts, Signals, and Side Effects
It’s been a crazy year since we were together for BLAM2025. We saw multiple releases of new foundation models that fundamentally shifted people’s perception of what AI could accomplish. December 2025 marked a turning point in how AI models could code with limited supervision. This shift has revolutionized professional programming and empowered non-professional developers to build everything from working prototypes to complete applications.
Simon Willison (who we featured last year) has a great interview on the Lenny Podcast last week which provides a balanced view of the transformations in the models with a focus on how it’s impacting coding. It’s well worth a listen, but you can also find the transcript on the site next to the share buttons at the top, however, we recommend listening if you can. Even if you aren’t interested in code development, we think you’ll still find this interview interesting but you might consider fast forwarding from around the 40min to when he begins to talk about AI and Security at 1hr,16 min.
Simon’s interview provides a nice look back at the last year’s developments as well as where things are headed. He also touches on some of the societal impacts which we’ll also be talking about this year on BLAM!.
For this week, our recommended reading is the cover story from the Atlantic last month entitled “America Isn’t Ready for What AI Will do to Jobs” which looks at the evolving conversation around how AI might shape the employment landscape in the future. Writer Josh Tyrangiel talks to people on both sides of the debate and provides a balanced, though admittedly somewhat scary view of what might be to come (there’s an audio version too!)
This week, we encourage you to learn more about MOAT, the Multi-Office particle Accelerator Team and their efforts to use AI tools to improve the performance and operation of particle accelerators. You can read about MOAT on LBL Newscenter, as well as read about LBL’s predecessor efforts around the Accelerator Assistant and the Osprey on NVIDIA’s blog (these became part of MOAT).
Theme: Gemini and Google Workspace
Learn how Gemini is weaved throughout the Google Workspace products. Gemini functions as a real-time collaborator embedded directly into the tools you use every day, like Gmail and Google docs.
Before you do anything we recommend you register for Google Skills at go.lbl.gov/skills, Google’s learning platform of short educational videos on how to use their products. We’ll be pulling specific videos from this library throughout the BLAM series.
Resources:
Online Courses: Before you do anything we recommend you register for Google Skills at go.lbl.gov/skills, Google’s learning platform of short educational videos on how to use their products. We’ll be pulling specific videos from this library throughout the BLAM series.
For this week we recommend the following videos:
Introduction to Google Workspace with Gemini: This 10 minute video is a great intro video to the basics of AI defining what generative AI is and
If you want deeper dives into Gemini in each of the following Google Products, we recommend the following:
Introduction to Gemini in Gmail (2 minutes)
Introduction to Gemini in Docs (5 minutes)
Introduction to Gemini in Sheets (5 minutes)
This week’s webinar: Wednesday, April 8th at noon. Gemini and AI Workspace with Adam Stone, where he’ll walk you through examples of how to use Gemini across the different Google Workspace products.
Office Hours: Thursday, April 9 at 9:30AM with the Collaboration team
Try using Gemini in one of your Google platforms and share your experience in the BLAM! Google Chat. Some examples of different uses could be:
Google Docs: have Gemini summarize a document you drafted for Elements article
Google Docs: have Gemini proofread your document
Google Sheets: use the Ask Gemini sidebar to do some data analysis: by asking, "What are the top 3 trends in this sales data?" or "Summarize the outliers in column C."
Google Chat: Use Gemini to create a list of action items from Google Chat conversation
Google Gmail: use Gemini to summarize Gmail thread to top points
This week's theme: Deep Research
This week we’ll start AI for Science with a deep dive into Deep Research tools powered by AI. These tools are transforming scientific discovery and knowledge synthesis. Participants will get hands-on exposure to leading platforms including Gemini Deep Research, ChatGPT Pro, Google Co-Scientist, AllenAI Astra, and more to learn how these systems can accelerate literature review, generate hypotheses, analyze evidence, and surface insights across large bodies of research.
This first week we'll focus on ways to help researchers understand how AI-assisted deep research can augment scientific workflows and support faster, more informed discovery.
Resources - Check out the different AI platforms, which will be the focus on this week’s 201 webinar:
Gemini Deep Research: Gemini Deep Research is an AI-powered agent within Google Gemini that automates comprehensive, multi-step research by browsing hundreds of web sources, Google Drive, Gmail, and Docs to create detailed reports in minutes. It features automated planning, reasoning, and, with the Google Gemini 3 model, provides enhanced, cited, and interactive analysis.
CBorg Deep Research: Berkeley Lab's multi-model AI platform, CBorg, to perform advanced, multi-step scientific research tasks.
Valency Research Assistant via CBorg: Agent that uses Valency Research MCP for its AI
ChatGPT Pro: ChatGPT Pro is a premium subscription tier of ChatGPT designed for users who need higher performance, greater usage limits, and access to the most advanced AI models and tools.
Google Co-Scientist: is a Gemini 2.0-powered multi-agent AI system designed to accelerate scientific research by generating, critiquing, and refining research hypotheses. It operates as a virtual partner that analyzes literature to produce novel, evidence-based research proposals and experimental protocols.:
AllenAI Astra: Asta (often referred to as Asta by Ai2) is an open-source, agentic AI platform developed by the Allen Institute for AI (Ai2) designed specifically to accelerate scientific discovery. It acts as an intelligent research assistant, assisting scientists with tasks such as literature review, data analysis, and information synthesis.
Futurehouse: Futurehouse accelerates scientific discovery is aimed at automating high-level research tasks like literature review, chemical synthesis, and experiment design, aiming to complete work in minutes that takes humans months.
Elicit: Elicit AI is an AI-powered research assistant developed by Ought that automates literature reviews and data extraction for researchers. It uses language models to search millions of academic papers, summarizing key findings and extracting information into structured tables without requiring perfect keyword matches.
Scite.ai: Scite.ai is an AI-powered research platform that helps users discover and evaluate scientific literature by analyzing over 1.3 billion citations to classify them as supporting, mentioning, or contrasting the claims of a paper. It features an AI Assistant for conversational searching, a Reference Check tool, and a browser extension, allowing researchers to quickly verify the credibility of studies.
SciSpace is an AI-powered research platform designed to streamline the academic literature review and writing process. It functions as a comprehensive, "all-in-one" AI Research Agent, allowing users to find, understand, and analyze scientific papers, manage citations, and generate content from a database of over 200 million academic articles and 50 million open-access full-text PDFs.
This Week’s Webinar - Wednesday, April 8th at 1:30PM: Andrew Schmeder, ScienceIT consultant will discuss AI platforms for Deep research including, Gemini Deep Research, Google Go-Scientist, and more
Office Hours: Thursday, April 9 at 9:30AM with the ScienceIT team
Select two of the deep research tools discussed in the class and give them the same prompt, on a scientific topic or question you are familiar with. Share your opinions about the pros and cons of each result with the Google Chat Group for 201.
Theme: Building foundations for AI-Assisted Programming
BLAM Coder Track kicks off in Week 1 with a focus on building the foundations for AI-assisted programming. Participants will set up their development environment and learn how to connect to the CBORG API, enabling access to advanced AI capabilities within their coding workflows.
The week will also feature a hands-on day exploring tools such as Claude Code and the Root Code extension, demonstrating how modern AI coding assistants can accelerate development, improve productivity, and support more efficient experimentation.
This week’s Webinar - Wednesday, April 8th at 3:00PM: Join Tim Fong, ScienceIT consultant, to discuss and get a walk through of AI-assisted programming with Claude Code and RootCode extension. from 4PM onward are Office Hours for the BLAM Coder group to ask questions
Try using one of the AI platforms discussed in this week’s webinar and discuss in the Google Chat for 201.