Hi, I'm Forrest

Category: Wondering

  • What If Goals Drive Everything?

    May 21, 2025

    I’ve been thinking about this for weeks and I think I finally figured out the missing piece.

    Remember when I talked about the trigger-action loop? Trigger → Cognitive Activation → Appraisal → Action → Outcome?

    I said there was no reward. That the habit loop has Cue → Routine → Reward, but our loop didn’t have that third piece.

    Well, I was wrong. The reward is there. It’s just not where I thought it was.

    The reward is the goal.

    Goals Give Purpose to Everything

    Every interaction you have is driven by an underlying goal. Not like “I want to make a million dollars” goal. More like “I want this person to trust me” or “I want to reduce my uncertainty about this situation” or “I want to maintain my professional standing.”

    These goals are what give the entire loop its purpose.

    Example

    You’re in a meeting with your boss. Your brain isn’t just processing information. It’s running toward a goal:

    • “I want them to see me as competent”
    • “I want to avoid looking stupid”
    • “I want to get approval for my project”

    Every word you say, every decision you make, is in service of that goal. And when you achieve it? That’s the reward. That feeling of “okay, that went well.”

    But here’s what’s wild—you don’t consciously think about these goals most of the time. They’re just running in the background. Guiding everything you do.

    The AI Problem

    And AI agents? They don’t have goals. They have tasks. They have instructions. But they don’t have that underlying motivation that shapes every decision.

    This is why AI feels robotic. Because it’s not goal-seeking. It’s instruction-following.

    What I’m Building

    An AI that has goals. Not just “complete this task” but “maintain this relationship” and “reduce uncertainty about this situation” and “build trust with this person.”

    Goals that persist across interactions. Goals that shape behavior. Goals that create the “reward” signal when they’re achieved.

    Because at the end of the day, every interaction is a goal-seeking mechanism.

    And if your AI doesn’t have goals, it’s not actually thinking. It’s just executing.

    What goals does your AI have? Or does it even have any?

  • How Expectations Form

    April 16, 2025

    I’ve been obsessed with expectations lately.

    Not the “I expect you to do better” kind. The cognitive kind. The automatic predictions your brain makes about what’s going to happen next.

    Because here’s what’s wild—your brain is constantly predicting the future. Constantly. And you don’t even realize it’s happening.

    Expectations Are Predictions

    Here’s what I realized: Expectations are predictions based on beliefs.

    When you walk into a meeting with Client A, your brain automatically predicts:

    • How they’ll greet you
    • What they’ll want to talk about
    • How long the meeting will take
    • Whether they’ll be happy or frustrated

    You don’t consciously think through these predictions. They just happen. Based on your accumulated beliefs about this client.

    And here’s the key: What happens next is measured against that expectation.

    If Client A is friendlier than expected → Positive surprise → Updates your belief about them

    If Client A is more frustrated than expected → Negative surprise → Updates your belief about them

    If Client A acts exactly as expected → Confirmation → Strengthens your existing belief

    This Is How Humans Learn

    Through expectation violations.

    But AI agents? They don’t form expectations. So they can’t learn from violations.

    Every interaction is equally surprising (or equally unsurprising). There’s no baseline to measure against.

    Example

    You tell an AI: “Client A usually responds within an hour.”

    Next day, Client A takes 6 hours to respond.

    A human would think: “Hmm, that’s unusual. Something might be wrong. I should check in.”

    An AI would think: “Client A responded.” (No expectation. No violation. No learning.)

    This is the missing piece.

    What I’m Building

    An AI that forms expectations based on its beliefs, then updates those beliefs based on whether reality matched the expectation.

    The AI learns:

    • “I expected Client A to respond in 1 hour (based on 10 past interactions)”
    • “They actually responded in 6 hours”
    • “This violates my expectation”
    • “Either my belief was wrong, or something unusual is happening”

    Over time, the AI gets better at predicting. Not because I programmed every scenario, but because it’s learning from expectation violations.

    The Difference Between Novice and Expert

    The difference between a novice and an expert isn’t knowledge. It’s the quality of their expectations.

    Experts know what “normal” looks like. So they notice immediately when something’s off.

    AI should work the same way.

    If your AI doesn’t form expectations, it can’t learn from surprises. And surprises are where the learning happens.

    How does your AI handle unexpected outcomes? Does it even know what “unexpected” means?

  • What Is a Belief, Really?

    February 19, 2025

    I’ve been thinking about beliefs lately. Not the philosophical kind—the cognitive kind.

    What actually is a belief?

    Most people would say “something you think is true.” But that’s too simple. That doesn’t explain why some beliefs are rock-solid and others change with one conversation.

    Why do I believe some things so strongly I’d bet money on them, and other things I’m like “yeah, maybe, I guess”?

    Beliefs as Compositions

    I’ve been wrestling with this for weeks and I think I’m onto something: What if a belief is actually a composition?

    It’s built from three things:

    1. Assumptions – Things you believe based on your life experience and common sense
    2. Opinions – Assumptions influenced by outside information (other people, research, news)
    3. Experience – Immutable things you’ve personally witnessed

    Here’s where it gets interesting: beliefs have different weights depending on what they’re built from.

    Examples of Belief Strength

    Weak belief: “I assume Client A prefers email over Slack”

    • Built from: One assumption
    • Strength: Low
    • Changes easily: Yes

    Medium belief: “I believe Client A prefers detailed reports”

    • Built from: Two assumptions + one opinion (their boss mentioned it)
    • Strength: Medium
    • Changes with: Contradictory evidence

    Strong belief: “I know Client A gets frustrated when I’m late”

    • Built from: Three direct experiences where they said so
    • Strength: High
    • Changes with: Sustained contradictory pattern

    This is how human cognition actually works. We don’t treat all beliefs equally. We weight them based on the evidence that supports them.

    The AI Problem

    But AI agents? They treat everything as equally true or equally uncertain.

    An AI can’t tell the difference between:

    • “I think this might work” (weak assumption)
    • “I’ve seen this work 10 times” (strong experience-based belief)

    Moving Forward

    If we want AI that learns like humans, we need to model beliefs as compositions, not binary true/false statements.

    The question isn’t “Does the AI believe X?”

    The question is “How strongly does the AI believe X, and what evidence supports it?”

    That’s the foundation of everything I’m building.

    How do you think about belief strength in your AI systems? Or do you even track it?