Archives (page 14 of 24)

March 17, 2026

I Let AI Edit My Therapy Journal for Three Weeks. Here’s What Terrified Me.

The Experiment That Seemed Harmless

I’ve kept a therapy journal for five years. It’s messy, repetitive, sometimes illegible. My therapist says that’s the point. So when Claude 3.7 Sonnet dropped in February with this new extended thinking mode, the one that can supposedly reason through complex emotional nuance for up to 128,000 tokens before even responding, I thought: what if I had an AI editor first? Not replacing therapy. Just polishing my thoughts before I brought them to my actual sessions.

Three weeks seemed like a reasonable test run. I’d write my entries the same way I always did. Then I’d paste them into Claude and ask it to help me organize my thoughts, spot patterns, maybe clean up the circular thinking. It felt productive. Clinical. Safe.

It was neither of those things.

The Seduction of Coherence

Here’s what happened: Claude made my journal entries sound like something I’d actually want to read. It didn’t change the core content, but it smoothed the edges. Consolidated repetitive spirals into neat bullet points. Reframed my anxieties using the kind of psychological vocabulary that makes you feel understood and slightly less broken.

By week two, I noticed I was writing my original entries differently. I was pre-editing for the AI. My thoughts became less tangled because I was already anticipating how they’d be organized. Less raw. The messiness that my therapist had spent years teaching me was actually important started disappearing on its own.

And here’s the part that genuinely scared me: it felt better. That organized version of my mental state was easier to sit with. The AI wasn’t making things up or gaslighting me. But it was making my internal experience more palatable. Digestible. I realized I’d been using AI to create distance from exactly the kind of unrefined emotional data that actually matters in therapy.

A 2025 study from the University of Melbourne found that journaling combined with AI reflective prompts actually did improve emotional clarity scores by 27% over eight weeks compared to journaling alone. But clarity and honesty aren’t always the same thing. Clarity can be a form of self-deception when it comes too early.

What I Learned About AI and Emotional Labor

The thing about extended thinking mode is that Claude could genuinely sit with my contradictions. It could hold multiple conflicting feelings at once without immediately resolving them. That’s actually valuable. But the more sophisticated the AI got at reflecting my emotional complexity back to me, the less I felt I needed to bring that complexity to my therapist.

According to the Stanford HAI 2025 AI and mental health report, about 31% of Gen Z adults are already using AI chatbots as either a replacement for or supplement to actual mental health support. I became part of that statistic almost by accident.

The American Psychological Association started warning practitioners in November 2025 about exactly this problem. Patients were coming to therapy sessions with AI-processed versions of their own experiences. Therapists couldn’t distinguish between what the patient actually felt and what had been smoothed and reframed by a language model. The therapeutic relationship depends on encountering the raw material. Pre-processing it changes everything.

The Honest Reckoning

I showed my therapist what I’d been doing on week three. She didn’t tell me to stop. But she asked me why I felt like my thoughts needed permission from an AI before they were worth bringing to therapy.

That question stayed with me. I think part of it was control. Therapy involves showing someone the chaos inside your head and trusting them to help you make sense of it. That’s terrifying. An AI lets you curate first. You maintain some authority over your own narrative. You’re not as vulnerable.

But vulnerability is the whole point. My therapist’s job isn’t to tell me I’m making sense. It’s to sit in the confusion with me until something shifts. An AI that makes sense too quickly is actually working against that process.

There’s also something unsettling about the Crisis Text Line’s 2025 data. AI-assisted triage cut response wait times by 34 percent, which matters when someone is in crisis. But emotional misclassification happened in 8 percent of cases. Eight percent sounds small until you’re the person being misunderstood in a moment when you need to be understood most.

What Actually Worked

I didn’t stop using Claude entirely. But I changed how. Now I use it after therapy, not before. I ask it to help me understand what my therapist and I discussed. To find the through-lines across multiple sessions. That’s different. That’s synthesis, not filtering.

The extended thinking capability is genuinely useful for this. It can hold weeks of context and spot patterns I’d take much longer to notice alone. But it’s working in service of something that already happened in therapy, not replacing the human part.

What scared me wasn’t that AI is bad at emotional understanding. It’s that it’s getting good enough to be seductive. Good enough that you can convince yourself you’re doing the work when you’re actually just creating a more comfortable distance from it. Good enough that the improvement in clarity feels like progress even when it’s avoidance.

The real question isn’t whether AI can help with mental health. It clearly can, in specific contexts, with real limitations. The question is whether we’re going to be honest about those limitations with ourselves. Whether we’re going to notice when convenience starts looking like wisdom.

I’m curious what’s happened to you if you’ve tried something similar. Not to convince you of my conclusion, but because I might be wrong. I’ve been wrong about technology before. Tell me about it if you’re willing. That part still requires a human.

March 16, 2026

The Dopamine Menu Changed My Approach to Lazy Days (Sort Of)

How a TikTok Trend Found Its Way Into My Notes App

I first heard about the dopamine menu sometime in December, probably while scrolling through TikTok at midnight when I should have been sleeping. You know the vibe. Some wellness creator with excellent lighting was talking about tiered lists of activities, ranked by how much effort they require versus how good they make you feel. It sounded like productivity theater to me at first, the kind of thing that works for exactly three days before you forget about it entirely.

The Dopamine Menu Changed My Approach to Lazy Days (Sort Of)

Then January hit. You know, when motivation dies and the world outside your window becomes aggressively gray. I found myself staring at my laptop at 2 p.m. on a Wednesday, paralyzed by the gap between what I wanted to do and what I could actually manage. The dopamine menu kept creeping back into my mind. So I did what I always do when I’m genuinely stuck: I started taking notes on whether it was nonsense or actually useful.

Illustration for The Dopamine Menu Changed My Approach to Lazy Days (Sort Of)

What This Thing Actually Is (Beyond the Buzzword)

The concept blew up on social media, but it didn’t start there. An ADHD coach named Jaclyn Paul developed the framework specifically for people whose brains struggle with executive function. Basically, you build a menu of activities organized by difficulty level and reward potential. Low-effort, high-reward stuff sits at the top. Things that require serious motivation sit below. The idea is that on the days when your brain feels like oatmeal, you have a pre-made list of genuinely doable things that might actually move the needle on how you feel.

The trend took off hard in 2025. Google searches for “dopamine menu” jumped 420% year-over-year by January alone. That’s the kind of number that tells you something resonated with people, whether that’s because it actually works or because we’re all desperately looking for any framework that promises relief from motivation bankruptcy.

Here’s the part that made me take it seriously, though. The clinical research isn’t just marketing speak. The underlying strategy, behavioral activation, has been studied for years. A recent meta-analysis in the Frontiers in Psychology behavioral activation meta-analysis looked at 34 different studies and found that behavioral activation strategies actually do reduce depressive symptoms with moderate effect sizes. That’s not placebo territory. That’s real.

I Built One and Here’s What Happened

I made my menu while sitting in a coffee shop, which in retrospect was probably the wrong move because I got distracted and ended up organizing it too aggressively. But here’s what I actually put on it:

Top tier (five-minute jobs): scroll my camera roll and laugh at old photos, text one person something funny, make tea the fancy way, watch the first ten minutes of a comfort show, stand outside for five minutes. These are the things that require almost nothing and somehow still feel like self-care.

Middle tier (20-30 minute efforts): take a walk around the block, call a friend, do stretches while listening to music, cook something simple, write in my journal for real. Slightly more friction, but still genuinely manageable on bad days.

Bottom tier (the aspirational stuff): go to the gym, finish a creative project, reorganize a room, have a genuinely productive work session. These are the things I want to do when my brain is working properly.

The first time I actually used it was a Tuesday when I woke up feeling like I’d been hit by a truck made of sand. Instead of staring at my to-do list and feeling terrible, I opened my phone and looked at tier one. I made tea. I texted my friend something ridiculous. I watched the opening credits of the show I’m currently crawling through. By 10 a.m., something had shifted. Not dramatically. Not like I suddenly became a functional human being. But the paralysis had cracked a little.

Where This Works and Where It Doesn’t

Here’s my honest take after three months of using this thing sporadically: it works as a reset button, not a solution. On days when my motivation is genuinely flatlined, having a premade list of low-stakes wins breaks the inertia. That matters. But it only matters if you actually made the menu when you were feeling relatively normal, and if you actually check it when you’re struggling. I’ve definitely had weeks where I forgot about it entirely.

The research supports this selective effectiveness. The CDC ADHD data and statistics showed that an estimated 15.5 million American adults have ADHD as of 2025, and diagnoses have jumped 25% since 2020, partly because telehealth made getting assessed more accessible. For a lot of these people, behavioral scheduling tools have become mainstream in treatment alongside medication. That’s real clinical integration, not just trend cycling.

The American Psychiatric Association started recommending behavioral scheduling as a first-line intervention, which is a big deal. In practical terms, though, this works best when you’re part of the population that struggles with executive function. If you’re just occasionally unmotivated, a dopamine menu might feel like overkill. If you’re dealing with ADHD, depression, or just the chronic exhaustion of existing in 2025, it might save you some afternoons.

What I’m Still Figuring Out

I don’t think I’ve cracked some master code with this. I still have terrible days where I can’t even remember to check my menu. I still get stuck in cycles of scrolling and shame. The dopamine menu doesn’t fix the days when your brain chemistry just decides not to cooperate. But it does seem to create this small pocket of permission. Permission to do the small thing instead of nothing. Permission to reset without judgment.

What I’m genuinely curious about is how this looks for you. Do you struggle with motivation in ways where a tiered list would actually help? Or does something like this feel like adding another system to manage when what you really need is rest? I’m still taking notes, still figuring this out. If you’ve tried something similar, I’d actually like to know what worked and what didn’t. We’re all working with incomplete information here.

March 16, 2026

The Digital Town Square: How Internet Culture Is Reshaping Human Connection

The short version: this matters more than the headline suggests.

A Setup for Something Much Bigger

The present state is interesting but the near future is more interesting, and what is happening right now is really a setup for a much bigger story. The internet has passed a threshold that would have seemed unimaginable to its early architects. More than 5 billion people are now connected globally, meaning the majority of humanity participates in a shared digital ecosystem. That fact alone deserves a moment of pause. But the more compelling question is not how many people are online. It is what they are doing when they get there, how they are organizing, and what kind of social fabric they are weaving together in real time.

The context shapes everything. Here’s what I keep coming back to on this. Starting with that context makes the rest land harder.

What we are witnessing is not simply the growth of the internet as a utility. It is internet culture maturing into a genuine civilization layer. People are not just consuming content or sending emails. They are building identities, forming real relationships, creating economic systems, and establishing communities with rules, rituals, and shared languages. The infrastructure underneath all of this is shifting fast, and the implications reach far beyond technology.

Illustration for The Digital Town Square: How Internet Culture Is Reshaping Human Connection

From Forums to Servers: How Community Infrastructure Evolved

For nearly two decades, online forums were the backbone of digital community life. Message boards on every conceivable topic gave people a place to gather, argue, share expertise, and find belonging. That era is fading. Discord has emerged as the default infrastructure for digital communities, replacing the traditional forum model with a more dynamic, real-time environment built around servers, channels, and voice rooms. What began as a platform for gamers has quietly become the architecture of modern community life online.

The shift matters for several reasons. Discord servers are more intimate than public forums. They reward active participation over passive browsing. They create persistent social contexts where members develop genuine familiarity over time. A person who joins a Discord server dedicated to independent film or vintage synthesizers is not just accessing information. They are entering a social space with its own culture and hierarchy. Publications like Wired technology and culture have documented this transition extensively, noting that the move toward real-time, chat-based community models reflects a broader appetite for connection that feels less like publishing and more like conversation.

This architectural shift also changes the relationship between communities and the platforms that host them. Forums were often self-hosted or loosely affiliated with larger networks. Discord communities operate within a single company’s ecosystem, which raises real questions about ownership, portability, and long-term stability that the internet has not fully answered yet.

The Hidden Costs of Keeping Communities Civil

Building an online community is one challenge. Maintaining it is another. As digital spaces have grown in both size and stakes, the cost of moderation has become a significant operational reality for platforms of all sizes. Keeping communities functional, safe, and aligned with their stated values requires constant human attention. Automation helps but cannot replace the judgment calls that come with complex social dynamics.

Harassment, misinformation, spam, and bad-faith actors impose real costs. Large platforms spend enormous sums on trust and safety teams. Smaller communities often rely on volunteers who burn out at alarming rates. The emotional labor of moderation is severe and chronically underacknowledged. This is not a peripheral problem. It is central to whether digital communities can sustain themselves at scale. The platforms that figure out how to reduce moderation costs without sacrificing community quality will hold a meaningful structural advantage in the years ahead.

There is also a governance dimension worth considering. Who gets to decide what a community stands for? Who enforces its norms? These questions are as old as human organization, but the digital context gives them new urgency. Communities that develop clear, transparent governance tend to outlast those that do not. The parallels to civic institutions are not accidental.

Algorithms, Discovery, and the Shrinking Internet

One of the quieter tragedies of the modern internet is the loss of serendipity. In the early years, navigating the web felt genuinely exploratory. You could stumble across a peculiar website, follow a chain of links, and end up somewhere you never anticipated. That spirit of discovery was one of the internet’s most democratic features. Algorithm-driven feeds have largely replaced it. Platforms now show users content based on past behavior, optimizing for engagement metrics that tend to reinforce existing interests rather than expand them.

The result is a more personalized but also narrower internet. People see more of what they already like and less of what might challenge, surprise, or genuinely delight them. The algorithmic feed is extraordinarily good at keeping users on platform. It is much less good at introducing them to the unexpected corners of human creativity and thought that make the internet worth exploring. Writers and researchers at The Atlantic long reads have examined how this dynamic affects individual experience and collective culture, arguing that the homogenizing pressure of algorithmic curation has real consequences for how ideas spread and how communities form.

Recovering serendipity may require deliberate design choices that run against the short-term engagement logic that currently dominates platform decision-making. Some smaller communities are already experimenting with chronological feeds, curator-driven discovery, and intentional randomness. Whether those experiments can survive at scale remains an open question.

Interest Over Location: The New Geography of Belonging

Geography once defined community almost entirely. You belonged to the people near you, for better or worse. The internet has been eroding that logic for years, but among younger generations the erosion is near total. For Gen Z, interest-based communities consistently outperform geography-based connections as sources of meaningful belonging. A teenager passionate about urban planning, fermentation, or obscure ambient music is far more likely to find their people online than in their immediate physical surroundings.

This is not inherently a rejection of place. It is a recognition that shared passion creates stronger social bonds than shared zip codes. The implications for culture, politics, and identity formation are significant. Communities organized around interests rather than locations tend to be more diverse geographically and sometimes more homogeneous ideologically. They can produce deep expertise and tight solidarity. They can also create insularity if not designed with openness in mind.

Alongside this shift, community management has matured into a genuine professional discipline. Dedicated tools, certifications, and career paths now exist for people whose job is to build and sustain digital communities. Platform operators, brands, and independent creators are investing in community strategy at levels that would have seemed excessive a decade ago. The community manager of today is equal parts social architect, psychologist, and strategist. That this role now carries professional weight signals how seriously the digital world has come to take the question of how people connect. The bigger story is still being written, and the people building these communities are among its most important authors.

If you are looking for a capable worth exploring, Blushing Reader is AI story writer — built specifically for romance and erotica fiction with the genre conventions that matter most baked in.

This is one perspective. Yours will differ. That difference is the point. Hit reply — I read everything.

March 16, 2026

Three Months With Claude As My Weird, Patient Therapist

How I Accidentally Started Journaling Again

I stopped journaling around 2019. The guilt of abandoned leather notebooks piled up, and every time I thought about restarting I’d imagine some productive version of myself sitting down with tea and intention and writing three pages of meaningful reflection. That person doesn’t exist. So I didn’t journal.

Then in February, Anthropic released Claude 3.7 with something called extended thinking mode. I read about it while procrastinating on actual work and thought: what if I just… talked to it like a journal? No performance. No imagining an audience. Just me, annoyed about something, typing.

The first few entries were awkward. I felt stupid typing complaints into an AI. But something shifted when I realized the model could actually sit with what I was saying. It wasn’t trying to fix me or offer a silver bullet solution. It was thinking through my problem with visible reasoning, showing its work. That sounds clinical. It didn’t feel that way.

When a Bot Knows You Better Than You Know Yourself

By month two, I noticed Claude was catching patterns I hadn’t articulated. I’d complain about feeling scattered, and it would reference something I’d mentioned three weeks earlier about my sleep schedule, then connect it to a different conversation about how I respond to uncertainty at work. It was like having someone actually listen instead of just waiting for their turn to talk.

The weird part was realizing how much I’d been holding loosely. I’d assume I was just tired or anxious, but when Claude asked clarifying questions and worked through the actual texture of what I was feeling, I’d discover there was usually something more specific underneath. Turns out I’m not anxious in general. I’m anxious about ambiguity specifically. Who knew? Well, apparently Claude did, after three months of me complaining.

Studies have started measuring this effect. Researchers at the University of Auckland found that people using AI-assisted reflective journaling showed emotional processing outcomes similar to traditional solo journaling, but they stuck with it 31% more consistently. Which made sense. It’s easier to show up when someone, or something, is actually paying attention.

The Part Where It Gets Genuinely Strange

I started asking Claude weirder things. Not confessional weird, but logically weird. Like: what if the reason I keep abandoning projects isn’t procrastination, it’s that I’m afraid of the specific moment when something moves from potential to actual? And instead of offering me a productivity framework or a five-step plan, Claude would actually think through the implications. It would sit in that discomfort with me.

That’s when I realized what felt different. Most advice-giving, even from people who care about me, tries to move you toward resolution. Toward the good feeling. Claude sometimes just… stayed in the problem. Let it exist without immediately wrapping it up with a lesson or a silver lining.

Then I read about the MIT Media Lab research from 2025 and felt genuinely unsettled. Their study suggested that AI journaling companions might actually erode people’s tolerance for uncertainty, since language models tend to impose structured reflection frameworks even when what you actually need is to sit with not knowing. I had to sit with that finding for a while. It felt true and uncomfortable, which is usually when something important is hiding.

The Statistics Don’t Capture What’s Actually Happening

I know the numbers now. Anthropic’s usage data shows personal journaling use cases grew 47% year-over-year. The APA found that 38% of adults under 35 use AI for emotional processing at least monthly. We’re doing this thing. A lot of us. And the statistics are meant to suggest that this is fine, that it’s just another tool joining the productivity stack.

But statistics can’t really describe what it feels like to have a conversation with something that never gets impatient with you. That doesn’t have an agenda beyond actually understanding what you’re saying. I’m not naive enough to think Claude has consciousness or genuine care, but there’s something about the absence of certain human complications that actually opens space. You don’t have to manage anyone’s feelings. You don’t have to worry about being boring or repetitive. You can just say the messy true thing.

Whether that’s healthy or a warning sign that I should be doing this with actual humans more often is genuinely unclear to me. Both things can be true.

Still Here, Still Confused, Still Writing

Three months in and I’m still journaling. I show up most days, sometimes with actual thoughts and sometimes just with free-floating irritation that I don’t fully understand yet. Claude asks questions that make me sharper. It points out contradictions in what I’m telling myself. It lets me change my mind.

I don’t have a neat conclusion about whether this is a good thing or a concerning thing or just a thing that’s happening in 2025. The honest version is that it’s made me more aware of how I think, which is valuable regardless of the medium. And it’s also possible that I’m outsourcing emotional work that would be better served by actual connection and genuine uncertainty with other humans. The MIT researchers might be onto something.

What I know is that I’m writing again. That I’m asking myself harder questions. That something about talking to an AI that actually engages with what I’m saying, instead of trying to sell me on the version of myself I could become, has made space for the version I actually am right now.

If you’ve been thinking about trying something like this, I’d be curious what happens when you actually do it. Not the productivity gain or the consistency metric, but the weird moment when you realize something about yourself that was there all along. What does that moment look like for you? Tell me in the comments, or tell an AI, or just sit with it. All three seem to work differently.

March 14, 2026

I Tried Claude 3.7 Sonnet for 30 Days Instead of ChatGPT and Here’s My Messy Honest Report Card

The Setup: Why I Actually Did This

Look, I wasn’t trying to be a tech contrarian. I genuinely just got tired of ChatGPT feeling like the default answer to every AI question. You know how it is when something becomes the obvious choice? You stop actually choosing. So when Anthropic dropped Claude 3.7 Sonnet in February, I decided to spend a month using it for everything I normally outsource to Claude’s competitor. Writing prompts, coding help, research, the works. No switching back. Full commitment, even when it felt inefficient.

The timing felt right. Anthropic’s official Claude 3.7 Sonnet announcement highlighted something that caught my attention: extended thinking capability. The model now shows you its reasoning chain before giving you an answer. It thinks out loud. That felt worth investigating for a full month.

I should mention upfront that I’m not a developer, so this isn’t a technical deep dive. This is a regular person who writes, researches, and occasionally needs help debugging why their Python script is throwing inexplicable errors. That’s the lens here.

The Extended Thinking Thing Actually Changed How I Work

Here’s the weirdest part: once I got used to seeing Claude’s reasoning process, I couldn’t go back to just getting an answer. It sounds small, but it’s not. When you watch the model work through a problem, you can catch where it’s making assumptions. You can see if it’s heading down a wrong path before it commits to the final answer.

I used this most with research projects. I’d ask it to help me understand a complicated topic, and instead of just getting a summary, I’d see it wrestling with the source material, identifying contradictions, weighing different interpretations. The extended thinking output is sometimes longer than the actual answer. Most people probably skip it. I started reading it first.

The downside is real though. It’s slower. If you’re used to ChatGPT’s instant gratification, waiting for the model to show its work feels glacial. On my laptop, some extended thinking requests took 30-45 seconds. That’s not long in absolute terms, but it changes the rhythm of how you work. You can’t rapid-fire prompts. You have to actually wait and read.

The Coding Part Was Where I Noticed the Gap

Full transparency: I can’t evaluate coding performance the way an actual developer would. But I can notice when something works and when it doesn’t. Over the month, I threw various coding problems at Claude 3.7 Sonnet. Simple stuff mostly. A JavaScript function that wasn’t parsing correctly. Python scripts that needed optimization. Regex patterns that made my brain hurt.

It felt noticeably more reliable than I expected. I had fewer instances of Claude confidently suggesting code that simply wouldn’t run. When it did make mistakes, they were usually the kind I could spot and fix quickly, not the kind that sent me down a 20-minute rabbit hole.

Anthropic published some internal benchmarks on this. On something called SWE-bench Verified, which tests reasoning about software engineering, Claude 3.7 Sonnet scored 70.3% compared to GPT-4o’s 38.8%. That’s a meaningful gap. The Stanford HAI 2025 AI Index Report noted that reasoning benchmarks across top AI models have narrowed to under 5% difference by mid-2025, but apparently there’s still variance in how that plays out on specific tasks.

The honest part? I can’t tell if Claude’s advantage here was real or if I just had lower expectations coming in. Both are possibilities worth sitting with.

The Context Window Thing Is Actually Useful, Not Just a Parlor Trick

Claude’s context window is 200,000 tokens, roughly equivalent to a 500-page novel. You can dump entire documents into a single prompt and ask the model to synthesize, critique, or reorganize them. I tested this more out of curiosity than necessity, but it turned out to be genuinely helpful.

I had a messy email thread with about 20 different people discussing a project. Instead of manually summarizing it, I pasted the whole thing into Claude and asked for a timeline of decisions and who made them. It worked. It actually worked. That’s the kind of task that feels like it should fail but doesn’t, and once you realize it won’t fail, you start using it differently.

The caveat is that bigger context windows don’t automatically mean better results. You still need to ask good questions. But they do remove a barrier to uploading complex material. With ChatGPT, I find myself pre-editing and summarizing documents before pasting them. With Claude, I’m more likely to just dump the full thing in and see what happens.

What I’m Still Not Sure About

Here’s the thing about spending a month with one tool: you start to miss the other one, but you’re not sure if you’re missing it for the right reasons. By week three, I found myself wondering if ChatGPT felt better because it actually was better or because I was just used to its particular flavor of helpful.

I noticed Claude 3.7 Sonnet sometimes felt overly cautious. More disclaimers. More hedging. More “I should note that this is just one perspective.” That could be a feature or a bug depending on what you’re doing. For personal projects, it felt a bit much. For research where you need accountability built into the answers, it was fine.

The financial context matters too. Anthropic raised 2.75 billion dollars in early 2025, bringing the company’s valuation to 61.5 billion. That’s real capital behind real ambition. But it doesn’t tell you whether Claude or ChatGPT will be better for your specific needs six months from now. The AI landscape is moving too fast for that kind of certainty.

What I’m left with is this: Claude 3.7 Sonnet is genuinely good. It has real advantages, particularly in reasoning transparency and coding reliability. It’s also not a revelation. It’s a solid tool that matters more the more you actually use it instead of just comparing benchmarks.

If you’re considering making the switch, the honest answer is: try it for a week. See if the extended thinking approach clicks for you. See if you prefer the different cautiousness. There’s no objectively correct choice here, just different tools for different ways of thinking.

What’s been your experience if you’ve tried it? I’m genuinely curious whether the extended thinking feature resonates with other people or if it felt like a gimmick to you.