As part of the Human AIgency series, I’m exploring practical ways to leverage AI to become more agentic in our everyday lives. For this piece, I’m collaborating with my friend, , to spotlight how AI can support us in being more creative and experimental in the kitchen.

Sonja runs a Substack called and a virtual cookbook community called Color Club where she shares simple and delicious recipes and home cooking tips every week. Her intention is to educate us on the powerful effects (and the accessibility) of a colorful whole food diet.

We first met at Levels, the health tech startup on a mission to help people understand how their diet and lifestyle choices are impacting their metabolic health. We bonded over our shared obsessions for metabolically healthy food swaps, the perfect Notion setup, and the great outdoors.

I consider myself to be relatively novice in the kitchen. I’m a throw-everything-into-the-pan-and-make-a-stir-fry type of chef while Sonja is a home chef who creates new recipes on the weekly.

We chat about the fun ways we’ve used AI as our sous chef to level up our kitchen game.

AI in the kitchen

Cissy: I’m bullish on AI’s ability to help humans become more agentic, particularly when it comes to trying things we might typically shy away from.

In this case, I’ve historically spent very little time in the kitchen, partly because I’ve lacked the confidence to experiment. I’m perfectly content eating the same handful of meals on rotation so for most of my adult life, I’ve stuck to a few variations of the same recipes.

Using AI in the kitchen has allowed me to become far more experimental with mixing random ingredients together and more confident that whatever I’m cooking up will actually be edible. I’m cultivating a deeper intuition for cooking and feel more creatively empowered simply because I’m cooking more often.

As someone who loves cooking, what’s your stance on AI’s role in the kitchen?

Sonja: In the kitchen, I’ve basically replaced Google with ChatGPT for anything I’m wondering, want to learn more about, or want help brainstorming. I’ve become a better home chef by learning tips and techniques for things like quickly caramelizing onions (yes, it’s possible — see here!), braising, marinating, and more!

The thing I like about AI and cooking is that while it can help give ideas and techniques, AI can never fully replace the art of home cooking. It can help you determine what to make with what you’ve got in your fridge, what to do with random ingredients you have on hand, how to meal prep something new within the parameters of how you like to eat and cook, how to get the most flavor out of ingredients, or how to salvage a recipe gone wrong — but it’s still up to you to taste test and determine if you need more salt or spices and use your senses to modify dishes so they surprise and delight your taste buds. The best thing AI can do is help you (and home chefs everywhere!) develop a better intuition for how to combine ingredients and prepare meals in a way that helps you fall in love with cooking flavorful, nourishing meals at home.

The way I think about food is that it’s what helps us be and feel alive. Food meets our cellular needs, nourishes our bodies and minds, and supports our energy, health, and overall well-being. If ChatGPT can help me learn how to feel even more joy, pleasure from preparing and consuming naturally occurring unprocessed foods, then I’d say that’s a pretty excellent use case for AI!

ChatGPT: AI sous chef edition

Cissy: How do you leverage AI as part of your cooking process?

Sonja: I made a list and then ChatGPT helped me turn it into this epic table👇🏼🤣

The only thing ChatGPT can’t promise is that a recipe will taste good — that’s where a human absolutely must still be in the loop to test and modify. If I just whipped up recipes from ChatGPT and published them as is on my Substack or my Color Club, I’d have very few subscribers 🤣

Cissy: Two things I’ll add that have been particularly useful for my cooking process:

• Salvaging a recipe (e.g., replacing an ingredient if I’m out of it or messed up some measurement of the recipe)

• Taking a photo of what I have available and asking it to generate a good on-the-fly recipe that takes <X time

Cissy: Of all the ways you’ve used AI in the kitchen, do any stand out as favorites?

Sonja: I subscribe to an organic community sponsored agriculture (CSA) program near LA called Earth Matterz , which means every week I get a bunch of seasonal produce that I then have to figure out what to do with! I often turn to ChatGPT (see 2 min Loom below) to brainstorm what to do with it!

For example, this week, I got a Kabocha squash. It’s mid-March so I wanted something spring-y with it and I wanted it to be simple since I was going to make it during the week (think sheet pan or one pan meal). 👇🏼

Cissy: Where has AI fallen short as compared to a human recipe tester? What’s your process for modifying and improving recipes?

Sonja: AI has no palate (yet? 😱). It doesn’t know if a dressing has too much or not enough acid, or if it needs more fat to bring out the flavor. I’ve found when it comes to the right proportion of ingredients it’s often too conservative with salt, acid, and heat.

For example, for the squash miso sauce for this sheet pan dish — the first version ChatGPT gave me was SO boring and dull, it needed more lemon, more tahini, and a touch more sweetness. I experimented with 3 versions before I was happy with it.

Cissy: What’s a recipe you’re particularly proud of that you started by using AI?

Sonja: Here’s the Miso Kabocha Salmon with Asparagus and Snap Peas recipe I made this week with my ChatGPT sous chef!

I’ve got hundreds more recipes up on my Color Club, head here and use code MOREMYSELF for a free first month.

the future of AI in the kitchen

Cissy: What predictions do you have for how AI capabilities will evolve in the kitchen?

Sonja: While we’ve already seen robots in kitchens all around the country from Sweetgreen to White Castle, I predict we will see AI continue to show up when it comes to meal planning apps as well as even smart appliances (e.g., AI vision to check doneness and auto-adjust based on the recipe and real-time cooking feedback).

We’ll also probably see AI (if not happening already) being trained on massive datasets of cuisines and cultures to do flavor pairings and menu design. I can imagine AI helping usher in an even more personalized nutrition feedback loop: think eating a meal, automatically logging glucose, HRV, mood response, and refining future recommendations. It’s like Levels + a private chef!

However, I believe there is something so human and primal about cooking. I hope that AI never fully replaces the human experience of preparing food. When I prepare my own food, I feel a deeper connection to the food, the earth, and the natural world.

The only reason you are reading this sentence right now is because you eat food that fuels you and keeps you alive. Once you eat something, it literally becomes you and becomes part of your 37 trillion cells. One of my favorite quotes is by Wendell Berry.

Eaters, that is, must understand that eating takes place inescapably in the world, and that it is an inescapable and agricultural act, and that how we eat determines, to a considerable extend how the world is used. This is a simple way of describing a relationship that is inexpressibly complex.

— Wendell Berry, The Pleasures of Eating

No matter what happens with AI — positive or negative — we must always strive to not lose sight of the most important part: food is living fuel that connects us to the earth.

prompts for your AI sous chef

Cissy: What prompts would you recommend our readers try? I personally love:

• I’m baking X and Y — what should I season them with and how long do I need to bake them to make them flavorful and crispy?

• What’s a fun, cultural twist I could add to this dish (with Chinese flavor) while making it metabolically healthy?”

• I’m craving something comforting and colorful — can you help me create a bowl or plate that balances both using what I have in my fridge? (Add a photo)

Sonja: Any time you have any sort of cooking or meal planning question that pops up in your head like “I wonder what I should make this week” or “I wonder how I could use these two ingredients” — think of ChatGPT as you sous chef brainstorm partner.

Here are prompts I’ve used many times and love:

• I have X ,Y, Z and common pantry items — give me 5 ideas of dishes I could make in less than 30 minutes using whole food ingredients with about 30g of protein per serving

• I want to make a sauce that goes with XYZ — give me 5 ideas

• A recipe calls for X amounts of XYZ but I want to find a substitute, what can I substitute? What should I know about how the ingredients operate differently?

• Here are my ingredients I want to use and what I want to make. Write me a recipe for how to do it. (Then I test, modify, tweak, etc)

• What are tips to make AMAZING XYZ?

• I want to make X meals with seasonal ingredients. It’s X season and I live in X place. I avoid refined sugars, refined grains and seed oils, and try to make all blood sugar balanced meals. I like the cooking style of chefs like Sonja Manning, Molly Baz, Caro Chambers, Justine Dorion [fill in the name of your favorite chefs!]. Give me ideas for X meals!

If you enjoyed the recipe included above and want to try some of Sonja’s other recipes, head here, and use code MOREMYSELF for a free first month. (Not sponsored — I just love the work Sonja’s doing with Color Club and have personally enjoyed a number of her recipes!)

If you’re interested in other ways to use AI intentionally, you may enjoy:

If you’re new to either of our work, welcome! You can follow us here:

Cooking in Color

Colorful & easy recipes and bite-sized health tips for a metabolically healthy life.

By Sonja Manning

becoming | more myself

meditations on agency, humanity, and becoming more ourselves

By Cissy Hu

What are your favorite ways to use AI in the kitchen? Let us know in the comments!

Leave a comment

This essay is part of the Human AIgency series where I’m investigating how we become more human in the age of AI.

For years, I’ve been haunted by a recurring dream. Each time it played out, I’d wake up disoriented, grasping for meaning, wondering why this unsettling loop kept presenting itself. At first, I dismissed it as an ungrounded fear, but as the dream persisted, I began to wonder, what if this dream wasn’t random? What if it was somehow a missive of wisdom dispatched from the depths of my unconscious?

For centuries, dreams have been a source of mystery, interpreted as divine omens and psychological clues. Carl Jung, the pioneer of analytical psychology, believed that dreams are expressions of the unconscious — symbols revealing truths that we’re not yet ready to face. These messages arrive in our dreams in the form of archetypal symbols and personal associations. Our dreams, it turns out, are rarely as straightforward as they seem.

Historically, we’ve relied on intuition, myth, and psychology to decipher our dreams, but now, for the first time, technology offers us a new way to decode these messages. Artificial intelligence, with its ability to recognize patterns and draw connections between symbols, invites us into a dialogue with our unconscious selves. With the right approach, AI can serve as an interlocutor, helping us tune in more deeply to the parts of ourselves that linger just beneath the surface of our awareness.

In the dream, I found myself in the driver’s seat of a car, driving to an unknown destination. The landscape varied — sometimes, I drove along the roads of my hometown, routes I’d known for years. Other times, I’d find myself navigating an unfamiliar city. I was often in the car alone and always with an unsettled feeling that something was about to go wrong.

As the scene unfolded, it became increasingly challenging to control the car. Suddenly, right before the dream ended, the car would speed up. I’d lose control of the wheel as I exited a highway or drove toward a ditch on the side of the road that appeared moments before. The dream never played out long enough for me to experience the impact of the crash.

For years, this dream replayed over and over again — each time, I woke up unnerved by the recurring feeling of spinning out of control. Eventually, I chalked it up to a fear of driving, the result of years of city living where I’d stopped driving.

A few years ago, during a coaching session, my somatic coach offhandedly mentioned that she used to have a recurring dream of her mom erratically driving her car while she sat powerless in the passenger seat. Often, her dog would get thrown out of the backseat as her mom tore through the streets.

My jaw dropped as she shared what the dream had represented. Growing up, she often lived in fear of her mom, deferring to her mom for big and small decisions. Metaphorically, her mom sat in the driver seat of her life, calling the shots — and it didn’t make for a smooth ride. Her dog represented her natural instincts which rode backseat. The act of getting thrown from the car symbolized a loss of that protective intuition in moments of chaos.

Her story prompted me to explore the significance of my own recurring car dream. As I unraveled the dream, I learned that, in Jungian dream psychology, cars represent how we move through our (psychological) world. They often represent our ego and the conscious self that navigates daily life.

• Sitting in the driver’s seat meant that I felt in control of my own life

• Being alone in the car represented a feeling of independence or isolation

• A speeding car represented an anxiety of keeping pace or a fear of falling behind

• Losing control of the car right before an accident represented a fear of powerlessness as I watch life spiral out of control

As I recalled the times when this particular dream presented itself, it was often during a season of life when I was overwhelmed and felt things spinning out of control — whether I had a big decision to make or felt under water at work. Incredibly, all of these symbols reflected my state of mind with remarkable accuracy.

This was my first introduction to Jungian dream psychology.

The stories that emerge while we’re asleep often bear the precise wisdom we’re seeking in our waking lives, revealing parts of ourselves we may fear confronting. Rather than taking our dreams at face value, we can delve into their deeper layers to uncover their true significance.

Jung developed theories around archetypes that serve as universal symbols and patterns across cultures and time. Some common symbols that emerge in our dreams are our teeth falling out, encountering a snake, being unprepared for an exam or a meeting, showing up somewhere naked, or being chased.

The exploration in interpreting our dreams is becoming increasingly more accessible as new AI models are being released. With the right prompting, we can begin to translate the messages our unconscious minds are delivering us each night.

The process begins with recall — remembering details of our dreams: a skill we can cultivate through practice and attention, training ourselves to hold onto the storylines that emerged overnight before they slip into the abyss.

As we learn to orient ourselves to the inner workings of our dreams, we begin to identify recurring symbols and patterns. While some symbols may carry universal meaning, others hold meaningful personal significance. It’s in the space between universal archetypes and personal associations where meaningful dream interpretation lies — each dream is a conversation with our unconscious selves and an opportunity to transform cryptic messages into pearls of insight.

ChatGPT: pretend you’re a world-class Jungian dream psychologist

I came across Jungian dreamwork in 2023. Around this time, AI models and AI-powered apps were just starting to become widely available to the general public. By the end of the year, I’d logged my first few dreamwork interactions with ChatGPT.

Over the last year, I’ve refined my process, experimenting with different prompts and varying levels of specificity.

Recalling the message

Despite being an avid dreamer, the details of my dreams often disappear the moment I wake up. I’ll have a feeling that I’ve had a meaningful dream, but can’t quite put my finger on what it was. To help jog my memory, I started keeping a notebook by my bed to document the snippets I do remember in as much detail as possible. For example: this morning, I woke up confused about the dream I’d just had except I couldn’t actually remember what had transpired. The only thing I remembered was that there was a scene where I was changing the background on my phone over and over again.

Some questions to contemplate: What was the scene? Who is present? How do you feel? Are there any recurring patterns or symbols? How does the scene interact with what’s happening in your waking life?

Once I’ve noted the details I remember, I’ll engage in low stimulus activities like meditating, doing the dishes, or starting the laundry to create spaciousness for my mind to wander back to the dream on its own time. As soon as I started doing the dishes this morning, my dream from last night resurfaced as if it’d never left my consciousness.

I’ve found that my dreams are more potent the evenings following a bodywork or somatic coaching session — after I’ve consciously grappled with particularly emotionally charged content in my life. It’s my mind’s way of closing the loop on what I’ve drudged up and processing it at the unconscious level.

After reviewing the contents of your dream and reflecting on the emotions that came up, this is where ChatGPT comes in. Of the existing AI models, I’ve found ChatGPT to be the most comprehensive and nuanced in analyzing dreams.

Communicating the message

Curious about AI’s interpretative power, I started feeding my dreams into ChatGPT. At first, my prompts were simple requests to interpret my latest dream:

Over time, I found that the more context I can share on what’s happening in my life, the better the insights. I’ve come to prefer a stream-of-conscious account of the dream. Now, I typically use the voice function to capture the breadth of what unfolded. Voice dictation is particularly helpful when I feel unsettled by a dream or there’s a lot of nuance to capture. I’ve crafted standard ways to frame the ask, often sharing additional context around the circumstances I’m grappling with in my daily life. I remix prompts and will occasionally share snippets of longer journal entries to be analyzed.

Here’s a stream-of-conscious example from this morning:

And boy, did this morning’s analysis deliver.

If you’re relatively new to dreamwork, here are a few prompts to experiment with:

Pretend you’re a world-class Jungian dream psychologist. Please help me analyze and decode the dream I had last night. Provide at least three different interpretations of the dream, focusing on the mythopoetic meanings that aren’t obvious from a cursory read of the dream. Help me break down the different symbols in the dream with an eye toward more symbolic interpretation.

If you need more detail, ask me for more detail and share the reason you want those particular details.

I have a recurring dream where [xyz happens].

I feel [xyz emotion] during the dream, often unsettled by [xyz symbol].

As a Jungian dream analyst, can you help me break down its main symbols and emotions? What might this represent in my unconscious?

I had a dream where I spoke to [xyz person or figure]. As a Jungian dream psychologist, can you guide me through an active imagination exercise to interact with the figure I saw in the dream?

How can I integrate the wisdom of this dream into my relationships and daily life? What might this dream be asking me to do in my waking life?

Looking back at my recent dreams, are there any overarching themes or patterns I ought to focus on?

During the AI for Inner Work salon I recently co-hosted, someone raised a concern that we risk outsourcing our self-awareness when we rely on AI to deliver insights about our inner world rather than coming to those conclusions ourselves. It’s a valid point. But perhaps it’s not always about how we arrive at clarity, but simply that we arrive at it.

Using AI to explore our dreams isn’t about surrendering our intuition, it’s about sharpening it. The real work isn’t in the interpretation, but in discerning what resonates with us then transforming those insights into deeper self-awareness. It’s an act of summoning our agency to explore our unconscious mind.

For years, deciphering my dreams felt like navigating uncharted territory in the dark. AI hasn’t redrawn the map — it’s simply handed me a flashlight, revealing the paths that were always there, waiting to be traversed.

Subscribe now

I’d love to hear about the recurring dreams that visit you or ways that you’re using AI for dreamwork analysis. Let me know in the comments or say hi on Twitter :)

Leave a comment

If you enjoyed this piece, you may also like:

Thanks to Ryan for reviewing drafts of this essay.

Last fall, I returned to academia to study the psychology of human flourishing and agency. It had been 10 years since I graduated from college and I’d blissfully forgotten the intensity that comes with committing yourself to an academic institution. The force with which the program consumed me took me by surprise.

I kept pace with my prior publishing schedule through October, made some declarations about experiments I was planning to run with this Substack, then promptly got swept up by my scholarly pursuits. By the time November rolled around, the weekly theory papers I was writing and stacks of research papers I was pouring over caught up to me. I had no energy left to write this newsletter. I felt the gnawing discomfort of straying from my word and a desire to return to what felt comfortable — and yet, another part of me yearned to deepen my authority in a new frame of writing.

In writing this Substack, I’d found my way in crafting creative nonfiction and memoir narratives, but increasingly, I felt drawn to write with more intellectual rigor, exploring ideas that piqued my personal curiosity and were grounded in empirical research. One of the main reasons I’d decided to pursue a masters in applied positive psychology was to hone my ability to weave together the philosophical with the experiential and pressure test my ideas in an academic setting.

14 academic papers later, I’m ready to dig into the big questions I’ve been mulling over. As for school: with my maternal grandmother’s decline in health, I’m taking a leave of absence this spring to free up my time to travel between the US and China to be with her as she nears end of life.

With some spaciousness from Penn and renewed energy, my intention for writing this year is to harvest my learnings from the last semester, evolve my relationship with this Substack, and actually experiment with new forms of writing. In this iteration, think: more Aristotelian friends-type pieces.

The first experiment I’m excited to run is a series call Human AIgency, an investigation into human agency in the age of AI. I’m expecting this exploration to take shape in the form of essays, practical field notes, and conversations with builders and thinkers.

Agency is the embodied conviction that we have the power to create change in our lives and transform our visions into reality. In a paper examining agency through the lens of Greco-Roman philosophy, Martin Seligman — the father of positive psychology — argues that during periods throughout history when agency lies in the hands of gods, human progress stagnates. When agency extends from the divine to humankind, progress across the board — from economic to technological to political — ushers in eras of great flourishing. Individual agency is the engine that powers human progress, driving societies forward.

In the literature, the state of agency is defined as the amalgamation of three psychological conditions: self-efficacy, optimism, and imagination.

• Self-efficacy: the belief we can achieve a specific goal now

• Optimism: the belief that we can achieve that goal in the future

• Imagination: the belief we can achieve a variety of goals we set our mind to

After studying the arc of human flourishing and agency through millennia — from ancient philosophical texts to modern psychology research, I’m wrestling with these pressing questions:

• What does it mean to live well and become more human in the age of AI? How do we do it?

• How does the rise of AI impact our sense of agency? In what ways can we become more agentic?

• Will AI free us up to self-actualize or edge us closer to a full-blown meaning crisis?

• How do our relationships with AI bode for our human relationships?

• Can we use AI to become more emotionally and socially intelligent?

These questions were at the heart of an AI for Inner Work salon I co-hosted last week. As the group shared notes on their favorite ways to leverage AI practically, we covered a wide swath of use cases — from using voice AIs to get closure after a messy breakup to making sense of unconscious thoughts by analyzing patterns in recurring dreams to sharing years of journal entries to converse with our younger selves. And that was just in the emotional work space.

The pace of change in AI is accelerating at dizzying speeds. Just two years ago, interacting with OpenAI’s ChatGPT felt revolutionary — like a glimpse into the future. Now, it’s just one among a growing constellation of advanced models — Anthropic’s Claude, Google’s Gemini, Meta’s Llama, and X’s Grok — redefining the boundaries of possibility. These AIs paint new worlds with simple text prompts, solve PhD-level problems with the flick of a wrist, and autonomously navigate the streets of San Francisco. Each new release is more groundbreaking than the last.

Earlier this week, Dario Amodei, CEO of Anthropic gestured toward our not-too-distant arrival at a reality where AI capabilities surpass human abilities across nearly every domain. The future is unfolding before our eyes.

In this moment of time, we’re edging toward a new manifestation of the divine — an intelligence that approaches omniscience and omnipotence, not through supernatural creation, but through the technological evolution of knowledge and capability.

If history were to repeat itself, our greatest challenge will be finding the right balance between human and artificial (i.e., divine) intelligence. Without the proper safeguards, we risk not just a conscious delegation of agency following in the steps of ancient civilizations and their gods, but an unconscious surrender to a technology whose power we can’t fully comprehend. Unlike the gradual and incremental shifts in agency of the past, the relationship humanity forms with AI can reshape human behavior within a single generation.

To subvert this risk, it’s incumbent on us to take a step back and reflect on the humans we want to be and the lives we want to lead in this era, deliberately directing AI to free us up to engage with our art in the broadest sense and cultivate rich lives.

As AI becomes increasingly central to our daily lives, Aristotle’s notion of phronesis, or practical wisdom, serves as a compass in guiding us toward harnessing our inner wisdom. Phronesis — the unifying virtue that coordinates and directs all other virtues — empowers us in discerning the right decision, at the right time, for the right reasons. In moments when it’s tempting to outsource our intuition to a sycophant AI, eager to please and reinforce our biases, phronesis calls us back inward, urging us to reclaim our agency.

I’m excited to experiment with this series in pursuit of helping us become more human in an increasingly AI-reliant world and finding ways to bridge the philosophical with the practical.

I’d love to hear what questions you’re grappling with or the ways you’re using AI in your daily life in the comments or say hi on Twitter!

Leave a comment

If this series resonates, you may enjoy some prior musings on our relationship with AI & technology:

Or consider sharing this newsletter with a friend :)

Share

Thanks to Ryan, , and for their feedback on drafts of this essay.