Learning the Basics of AI

Learning the Basics of AI

This might be one of those posts that makes me seem like an old man. We’ll see. ??

For much of the last year, I’ve been thinking about how to most effectively build things using AI, and I’ve documented a bunch of those thoughts here. It turns out to be both hard and easy: easy to get something to happen, since the world of LLMs is really capable in some ways and moving fast. But hard to build satisfying, high-leverage experiences.

We are clearly still missing some basic reflexes and best practices. Memory and agency are on everyone’s mind now, and have been for a while, and yet, we still don’t have great consensus on what the right approaches are. Some companies are starting to deploy larger solutions in successful ways, and are beginning to confront issues like regression testing, cost optimization, and so on. Much of this is still happening by hand or experimentally in individual teams.

In a conversation this week, this reminded me a bit of the very early days of GUIs. I remember distinctly the process of learning, along with much of the industry, about basic capabilities that we so much take for granted that to mention them now does make me seem weirdly out of touch. Things like what a clipboard is, and how you interact with it, or how windowing works and the role of graphics masks, or even just the basic idea of a graphics library and the primitives it implemented (which evolved significantly over time as the hardware supported more and more sophisticated modes).

For a while, many of these things were issues you had to confront as a developer - you had choices, and you also had to implement much of the solution. Gradually, the approaches that work became clearer and were integrated into systems and libraries. Now, to mention some of these things seems painfully out of touch - they just work.

That process took many years for the desktop and then mobile and then internet worlds. It’s not clear how long it will take for the AI world, but it’s clear that it’s beginning. We will need to have something that feels like an OS for AI, that can identify and implement common patterns - some form of memory that enables agentic behavior perhaps, fundamental models for protecting and segregating user data, regression testing, behavioral monitoring, optimization and model selection, fine tuning…I don’t think we even fully understand what the problems are yet, much less the patterns and primitives.

In the desktop, internet, and mobile worlds, this progressed largely because of a desire to build more capable solutions. That’s likely the energy here as well, and having that discontent and ambition is probably the single thing we can each do to help push the field forward. Look for the problems that are in the way of building better and more valuable things. Think in terms of platforms and general patterns and build repeatable solutions where you can. Be ready to discard them when more information is revealed, or new capabilities appear.

As a young programmer, being able to draw pixels on the screen was magical, but limited, until we as an industry started to understand and build out these more common patterns and practices. I suspect the same needs to happen with AI.





This week I got an email from our car dealer asking if I wanted to talk about the additional services recommended for our car. I emailed back saying we were on a service plan and I did not know of any service recommendations. A return letter informed me that there was no additional service required. There was an explanation: "Sara Roberts was an AI. Sara got confused." This sort of problem is showing up more and more, and not necessarily on automated systems that anyone calls "AI." ??

回复
Robert Mowery (羅伯特·莫爾里)

激流回旋的行政领导 | 专注于创新、新兴技术和 Google Cloud| 早期投资者 | 海军老兵 | 可持续农民

10 个月

The GUI days, the Tools that made things easy and streamlined like Dan Bricklin's Demo Program & tools, Hypercard, Toolbook, then Marc Canter's MacroMind, all were leaps to simplify complex assembly if the underlying system and allow developers to create faster and better. Perhaps we will see tools enabling this for AI, AI agents, etc. that are operationally simple - allowing quick and simple connections of AI's to interfaces of the future/user choice, whether those are avatars, mixed reality, EMOTIV , Neruolink, Robots, or something we have yet to invent. Great thoughts and observations Sam Schillace.

回复

This is so spot on. The more I implement AI solutions and work in AI experiences the more I see the same pattern you are describing as an industry need. We are currently in the trenches of figuring our these patterns and platforms and building repeteable solutions within a specific context (typically either products or enterprise), and yet for AI to really shine on value and UX, we need that industry level standard so we can elevate all AI solutions.

回复
parth ??

building a thousand agents

10 个月

I wonder a lot if interacting with AI might feel more native in a realtime experience more like an RTS game, ie. rimworld, age of empires, civilization, starcraft... as opposed to pure chat. It's leading me to think more about game engines and agents that can use them.

要查看或添加评论,请登录

Sam Schillace的更多文章

  • AI analogies and historical lessons

    AI analogies and historical lessons

    How to make sense of it all. I've decided to keep the primary posts over on Substack.

    1 条评论
  • Motion, Thought, Systems and AI

    Motion, Thought, Systems and AI

    In which I ponder how motion is like thought, why LLMs are like early steam engines (hitting things and pumping water),…

    4 条评论
  • Looking back at the Schillace "laws"

    Looking back at the Schillace "laws"

    Way back in, I think, March or so of 2023, after I’d spent a little while trying to build things with GPT-4, I wrote…

    5 条评论
  • A strange tech parable

    A strange tech parable

    In my role at Microsoft, part of what I do is spend time with the leadership team that runs M365, Office, and Windows…

    12 条评论
  • Simplicity, Clarity, Humility

    Simplicity, Clarity, Humility

    There is an old joke: “sorry this letter is so long, I didn’t have time to write a shorter one”. It’s funny, but it’s…

    4 条评论
  • A matter of context

    A matter of context

    It’s interesting that, as we talk about using AI more and more, the phrase we use is “human in the loop” instead of “AI…

    3 条评论
  • The tension between Chaos and Order

    The tension between Chaos and Order

    I’ve been spending the last week in Japan, meeting with makers and crafts people. as always, it’s a humbling…

    4 条评论
  • No Prize for Pessimism

    No Prize for Pessimism

    A book! I’ve been writing these letters for about 12 years now. I started writing them when I was at Box, as a way to…

    10 条评论
  • Adding Value in the Age of AI

    Adding Value in the Age of AI

    If you wrote out all possible combinations of, say, 1000 letters, the vast number of them would be nonsense. And the…

    3 条评论
  • Don't use AI to make work for humans

    Don't use AI to make work for humans

    I’ve started to notice an interesting pattern. More enlightened teams and people are using AI to get lots of work done…

    5 条评论

社区洞察

其他会员也浏览了