Shaun Abram
Technology and Leadership Blog
Subscribe to RSS FeedI’m Begging You: Never Write With A.I., by Bret Stephens in the NY Times
A while ago, I did a quick blog post on my own personal approach to using AI when writing, i.e. don’t start with AI. Instead, start with a blank piece of paper, and while it can be beneficial to use AI to challenge your ideas, don’t copy and paste from it, lest you lose your voice.
AI is always tempting. It can quickly create comprehensive documents. Sometimes too comprehensive. It can be verbose, repetitive, bland. Content that looks contextually appropriate, with the right words in the right order, but adds little actual value. Fluff, you might say. Or AI slop, perhaps, which wikipedia defines as “digital content made with generative artificial intelligence that is perceived as lacking in effort, quality, or meaning”.
So I was intrigued to see this opinion piece in the NY Times (paywalled but trials available), by Bret Stephens, discussing the same topic.
Some highlights below, but I highly recommend reading the article.
Tags: AI, artificialintelligence, nytimes, writing
What is an AI agent?
TLDR: There iss still no standard definition for what an AI “agent” is, but one definition could be:
An AI agent is a LLM based tool capable of autonomous decision making by utilizing tools.
I discuss some of the common characteristics across the emerging definitions of agents and agentic behavior below.
Tags: agent, agentic, agenticcoding, AI, artificialintelligence
Research paper summary: Attention is all you need
Attention is all you need is an era defining paper from eight researchers who all worked for Google at the time. The title is a play on the Beatles song, Love is all you need.
The paper addresses some of the shortcomings of other common approaches in Machine Learning (such as recurrent and convolutional layers), in favor of a new architecture, the Transformer. It is based on an approach called self-attention where the model can associate the current token (or word) to any other token in the document, regardless of it’s position. This makes the model so much better at capturing associations and meaning across documents. Further, the Transformer architecture is able to to this processing in parallel, making it faster that preceding models.
I don’t have a Machine Learning background, so this paper was a tough read for me. My interpretation reflects that and I have relied on (and quoted from) other interpretations to help me, including:
- “The Illustrated Transformer” (github.io) by Jay Alammar, which is the most digestible interpretation I could find, and
- This summary from the University of Notre Dame (although exact authorship unclear)
Still, the original is short (16 pages) and is so influential that it was worth reading directly; it came to define the modern era of AI and of LLMs (Large Language Models).
Tags: AI, architecture, artificialintelligence, google, large-language-models, llm, machine-learning, neural-networks
Podcast review: Behind the Craft – Inside Ramp
I found the Behind the Craft: Inside Ramp (spotify.com) podcast an interesting watch. You can find it on Spotify or directly on the Behind the Craft site. The interview is with Geoff Charles (ramp.com), the CPO of Ramp, a fintech platform for corporate cards, and about how they are using AI to build.
I didn’t agree with some of what Mr Charles said, but it was thought provoking.
Some of the funny and/or controversial takes were:
- “a lot of PMS join product management because they might not be good enough at the engineering task”. Wow. That is a controversial opinion, especially from a CPO.
- Geoff suggested leaders should cancel 1-1s to focus on building their API skills. Personally I agree with leaders doing everything can to get up to speed with AI tools, including hands on experience, but not at the expense of neglecting your direct reports & team.
Other highlights…
Tags: AI, artificialintelligence, claude, podcast
My personal approach for writing with AI
My own personal notes on using AI when writing. AI is a great tool, but not one to be overused. Avoid creating AI slop. I put these notes together after reading Co-Intelligence: Living and Working with AI by Ethan Mollick, which is a good read.
Tags: AI, artificialintelligence
Book review: Co-Intelligence by Ethan Mollick
We covered Co-Intelligence: Living and Working with AI (amazon.com) by Ethan Mollick at a book club at work, and I found it to be a good read. Part 1 at least, part 2 perhaps less so.
Read on for a review and some notes.
Tags: AI, artificialintelligence, bookreviews, books
What are Small Language Models?
We hear a lot about Large Language Models (LLMs) such as OpenAI’s GPT (Generative Pre-trained Transformer), Google’s Gemini and Anthropic’s Claude.
Small Language Models (SLMs) are much less well known, but include the Phi family of SLMs from Microsoft, Mistral Small, and Llama from Meta.
The “large” or “small” in these models refers to both the number of parameters in the models, and the amount of data used for training. SLMs have both a smaller number of parameters and a smaller training data set when compared to LLMs.
Admittedly, the line between LLM and SLM is blurry. There is no well defined distinction between them, but below I cover some of the common characteristics.
Tags: AI, artificialintelligence, largelanguagemodel, llm, slm, smalllanguagemodel
Report Summary: 2025 DORA report
I’ve been a huge fan of the annual DORA (DevOps Research and Assessment) report since reading the Accelerate book that explains the DORA metrics and how to improve them.
The 2025 DORA report (pdf attached here) just came out and has a particularly interesting section on AI Capabilities, The DORA AI Capabilities Model. I summarize some of the points here that I found most interesting here, but the original is well worth a read.
Tags: AI, artificialintelligence, dora, dorareport, genai, summary
Blog post summary: The role of developer skills in agentic coding, by Birgitta Böckeler
Birgitta Böckeler, a Distinguished Engineer at Thoughtworks, recently posted a list of links that “describe what my colleagues and I are learning and thinking” around generative AI and LLMs. One in particular that I liked was “The role of developer skills in agentic coding” (written by Böckeler herself).
Böckeler states that agentic coding assistants are becoming ever more capable but, as the engineer, you still need to intervene, correct and steer. As well as detailing types of problems she sees with these tools, recommendations are provided on how to work around and mitigate.
(more…)Tags: agenticcoding, AI, artificialintelligence, summary
Summary: GenAI Perception vs Reality, from Jellyfish
I like this GenAI: Perception vs Reality pdf from Jellyfish. Short and sweet.
Jellyfish provides visibility into software development processes by integrating with Jira, GitHub, and CI/CD pipelines. As a result of their large customer base, Jellyfish are able to analyze more data than most individual companies can. In this case, they looked at 4,200 developers across 100+ companies, to see how using GitHub Copilot impacted productivity.
Some of their results include that engineers who use Copilot will have a 23% increase in coding speed and will ship 21% more. And that junior devs may have as much as a 4x increase in speed.
Tags: AI, artificialintelligence, copilot, genai, generativeai, github, githubcopilot, jellyfish
AI-First at Shopify; a memo from Tobi Lutke
CEO of Shopify, Tobi Lutke, posted an “internal” memo entitled “Reflexive AI usage is now a baseline expectation at Shopify“. It very publicly leaked and he has since shared it himself on twitter. It is generating a lot of discussion (e.g. CNBC, Forbes, Techcrunch).
Tags: AI, artificialintelligence, shopify, summary
Article summary: The 10x Developer vs. AI
The 10x Developer vs. AI: Will Tech’s Elite Coder Be Replaced? (thenewstack.io) is an article about embracing AI to create high-performing dev teams.
Tags: 10xdeveloper, AI, artificialintelligence