Voice of the DBA

Tag: AI

  • A Cloud Dependency Failure from Amazon

    I went to sleep while reading a Kindle book on my phone. I know because my hand dropped and the phone knocked me in the forehead. I set it on the nightstand and went to sleep again.

    I woke up and was planning on reading for a few minutes before starting my Monday. When I opened the Kindle app, it asked me to log in. When I did, Amazon said my account was locked and I needed to check an email. I did, and found one from Amazon.co.uk questioning a purchase I made and asking for me to very an order number and the last 2 digits of a card to charge.

    I wasn’t sure what to do, but since my wife was complaining about her Audible account, I knew I needed to do something.

    I didn’t like the email for multiple security reasons, including not giving me a way to verify the email online. Checking Amazon’s help page said I should have a link in the email to a form. Even when I called the Amazon help line, they said I should have a link when I logged in to do so. I didn’t on either the .com or .co.uk sites, but Amazon Customer Service eventually unlocked my account.

    I wrote about Dependencies in today’s editorial, inspired by this story and the Reddit story linked in there about losing access to Claude. That’s scrary, especially in this new automated world where AI agents might be examining activity and undertake this type of action to prevent other issues. It’s an overreaching type of judgment that starts to expose the dangers of a highly interconnected world.

    I get Amazon might want to ensure my purchase was legitimate and perhaps prevent future orders, but also locking my content away (books, movies, audio, etc.) because of a shopping issue seems extreme.

    It’s certainly my issue, and I’m not sure how I untangle things for the future, but it does make me think about limiting dependencies a bit more in the future. Or at least understanding the dangers of too much on one service.

    Whether that’s the books I read or the services my company gets from any one vendor.

  • Half of All Engineers

    The AI LLM boom seems to show no sign of slowing down. Each time I think we’ve reached some level of crazy use or predictions, things take another turn. I still find myself pinging back and forth between this will be amazingly good and horrifyingly bad.

    Sometimes on the same day.

    Today, I’m a little more down on AI. I was listening to Steve Yegge on the Pragmatic Engineer podcast, and they were discussing the curve of AI usage at companies. He points out that he’s mad that Amazon let 16,000 engineers go and might let more go. He worries that companies might let 50% of their engineers go. Not necessarily because the top 50% will be more productive with AI than 100% of engineers without it. Rather the concern is that companies will get rid of half their salaries to pay for the AI tokens for the other half.

    Steve Yegge is an accomplished software engineer that has worked at Amazon and Google. Steve wrote Gastown and has been someone who not only is successful at producing code but also thinks a lot about how we produce more software.

    How coordinated and powerful are the new models? Can they really do a lot of software work that we do today? Steve thinks so, but to be fair, he’s got a lot of experience and can architect and design software well, which means he can also guide AI LLMs and agents to write more code. He also thinks the latest models, like Opus 4.6, are way more capable that most people believe.

    I also caught this post on X about a paper predicting that AI might cause economic collapse as less knowledge workers are used in various tasks. This might happen faster than we can absorb those workers who are laid off in the name of AI back into the economy in other positions. It’s a scary thought.

    The positive side of this, at least from me, is that so many organizations move slower, and so many people aren’t extremely competent software engineers, so we’ll get a lot of bad software written by non-technical people that doesn’t scale. We’ll have more database (and application) performance issues, and that will slow the use of YOLO, vibe coding.

    Plus plenty of companies just aren’t implementing or looking to pay for lots of tokens. They’ll just move slower and the world will change, but not anywhere near the pace that Steve or others think it will. We’ll see, but let me know what you think.

    For a more positive spin, I’ve been reading Reshuffle, which is a little less depressing about the future.

    Steve Jones

    Listen to the podcast at Libsyn, Spotify, or iTunes.

    Note, podcasts are only available for a limited time online.

  • Questions and Answers from Running a Local LLM

    I had a few random questions from my Running a Local LLM on Your Laptop session at the Houston AI-lytics 2026 event last week, so this post looks at a few of those questions and my answers.

    Note: This stuff is changing rapidly, and there aren’t a lot of factual answers. A lot of what you should look for is guidance and rational reasons for leaning in some direction.

    Questions below:

    • Do we need an NPU? (Or what do I think of NPUs)
    • How do we audit or Test an AI LLM and know what is happening?
    • In which situations would you run a local model?
    • Which Model is Best?

    Do we need an NPU? (Or what do I think of NPUs)

    You don’t need an NPU to run a local LLM model, but they help with efficiency. An NPU is a Neural Processing Unit, which is a type of CPU that is designed to work with AI-type applications and process instructions more efficiently. This could be training a model or running LLM workloads.

    I think an NPU is a great idea for efficiency. We already know AI applications use a lot of compute and power. Just look at all the concerns over power/water and investments being made in new data centers for AI. Being more efficient helps.

    Just like a GPU helps with graphics and makes your laptop more efficient, an NPU will help, but it’s not required.

    How do we audit or Test an AI LLM and know what is happening?

    First, LLMs aren’t deterministic, so they might not return the same thing all the time. It’s hard to test a non-deterministic thing because we look to assert if a is passed in, b is returned. If I pass in a and sometimes get b, sometimes c, and rarely f, this is hard to test.

    I have no idea how to test a model for behavior in this case.You get useful results from experiments, and more often they are useful than un-useful to continue using it. If that happens, faster, then you have a better model. If it’s slower/more expensive/less useful, it’s a worse model.

    Auditing is looking at what happened, which means reaching into the processing of these GPT-type tools. There are some tools to help (AuditLLM), but I can’t speak to whether these are a) worth the effort, b) effective, or c) junk. I’m still learning here, too.

    In which situations would you run a local model?

    This is a hard one because there are a few situations in which I’d seriously consider a local model (including Amazon Bedrock/Azure AI/Google Vertex).

    First, when I’m worried about costs and I want to control them. While the vendors give you some limits and throttles, it can be expensive. In many cases, if I want to set controllable spend, a known spend, I might consider a local model in some service because I can allocate out capacity and know what is available, what it will cost, and who will be using it. Perhaps the cloud vendors will give us more controls and ensure we aren’t on “shared” systems, but any efficient use of hardware to do this will be for their benefit, not mine.

    Second, when I’m really concerned about data security. While most companies might promise they won’t use your data and will delete sessions, they might not, and they might make mistakes. And if they do, would they accidentally use my data, or send it in response to some sort of legal subpoena accidentally? If I’m outside the US or really worried, I’d run local models.

    Third, if I want to ensure that I have complete control over the training of the model or the prompts, I might use a local model where I know there aren’t any system prompts being injected into my context.

    Which Model is Best?

    Yes.

    There’s no good answer here. If you look at the list of models on Hugging Face, for example, there are lots and lots of models. None of us has time to test many of them, or even a small fraction. I think you have to depend on the community here to help you decide that any of these models are better for your situation.

    Think about what you want a model to do, what things are important to your problem space, and then look for a model that people think works well and does the type of things you want to do. Similar to how you interview a person for certain types of work, think about that for a model.

    The nice thing outside of the large LLMs is that you can use smaller models to fill in certain situations if you find you want to provide that capability a lot to your organization. I would see interpretation and linting of best practices in code, for example, using a smaller model that uses less compute, but it trained, or you fine-tune it for your particular situation (and save money).

  • Houston AI-Lytics 2026–Powerpoint Slides

    Thanks to everyone for attending my session on running a Local LLM.

    If you have any questions, please feel free to reach out with them. The slides with links embedded is below:

    Slides: Running a Local LLM.PPTX

    I’ll blog next week on a few of the questions people asked, so if you have anything you want answered, please reach out.