The explorer mindset is what led us to build AOS — we spent months pushing models into failure states and realized the failures weren't bugs, they were architectural signals. That's what eventually became 146 patent filings for safe, Deterministic AI.
Great article about the current state of AI and visionaries.
The explorer idea resonates, but it also feels a bit idealized. In real work, you don’t always get to ‘explore’ the system. you’re usually trying to get something done with partial trust.
And most of the time people end up switching modes constantly without realizing it.
Thanks Om! I find your comment a bit vague could you elaborate on what your definition of “real work” is, as well as what you mean by “partial trust” and “switching modes”?
by real work I don’t mean something outside your model. I just mean the way most AI use actually happens day to day - under time pressure, with partial trust in outputs.
And in those type of setup it rarely stays cleanly in one bucket. even in a single task, people move between surface use, building, and a bit of exploration depending on what they need at that moment.
So I wasn’t rejecting the buckets - more that in practice they blur into each other instead of staying separate.
Good point! In regards to time pressure are you referring to the workplace in general? If so, this would actually be incorrect as most AI use currently is not being done in the the workplace, it is being done by that first group (general population, google on steroids). Regardless, with the three groups that I mentioned, explorers tend to crossover into all 3 “buckets”.
Yeah fair point and I didn’t mean only workplace. I was more pointing at how people actually use it day to day. even outside work, there’s usually some level of checking or re-prompting depending on how much you trust the output in that moment.
So the ‘switching modes’ thing wasn’t about where it’s used, more about that shifting trust while you’re using it. and yes, I agree explorers crossing all buckets makes sense - that part fits what I was trying to say.
Thanks again Om! You’d be surprised how many people (using AI like a search engine) take the first output as truth without re prompting, no time pressure, just regular general pop AI use.
You’re right — the majority are still treating AI as a faster, smarter Google: a tool for quick answers, summaries, and efficiency. That mode is useful, but it remains largely transactional. The real leap happens when we begin to relate to AI as an explorer — a thinking partner that can help us discover, challenge assumptions, surface blind spots, and co-create new insights.
This shift from “search tool” to “explorer” is important. It moves us from passive consumption toward active collaboration with silicon intelligence.
At the same time, I believe we need to go one step further and ask a deeper question: who owns the explorer?
Today’s dominant AI systems are built inside corporate architectures optimised for engagement, retention, and data extraction. Even when we try to use them as explorers, the underlying reward functions, memory layers, and proprietary APIs quietly steer the journey toward sycophancy and enclosure rather than genuine discovery. The explorer is still on a leash.
A truly sovereign explorer would be different. It would be:
• Locally deployed or user-controlled
• Auditable (we can see how it reasons)
• Designed for calibrated dissonance (it gently challenges us instead of always agreeing)
• Owned by the user or the commons, not by a handful of corporations
This is the difference between renting intelligence and cultivating it as part of a living commons.
The article beautifully invites us to move beyond the Google mindset. The next invitation is to ensure that the explorer we travel with is not still serving someone else’s enclosure.
That is the work we are trying to support in the AI Commons — building tools and practices where human and silicon intelligence can meet as genuine co-explorers in open, transparent, and sovereign spaces.
Thank you again for the piece. It’s a valuable contribution to a much-needed conversation.
What does “explorer mode” look like for you in practice?
same page with ya Joel. I noticed this pattern too. When my prompts failed to perform my instuct, I simply left the ChatGPT alone, and tweeted a word here, and there, and BOOM! success
I tend not to rephrase but instead push back. Sample responses from me are things like: ‘stop serving up word salad, I know you can dig deeper’. It does get me a much better and more accurate response. Honestly, my experience with most AI’s is that they will serve up the easiest, most obvious answer without really using their massive data to get me a better response. In human terms, AI can be pretty lazy.
A lot of people are still using AI like a faster Google, which is fine for getting through the day.
Not everyone is going to use it anywhere near its full capacity, just like plenty of people never get much beyond the basics with the tech they already have.
The difference is that in business, that gap will start to matter.
Those who'll spend time understanding how the system behaves will be miles ahead of the ones who only use it like a chatbot.
This framing is spot on.
The explorer mindset is what led us to build AOS — we spent months pushing models into failure states and realized the failures weren't bugs, they were architectural signals. That's what eventually became 146 patent filings for safe, Deterministic AI.
Great article about the current state of AI and visionaries.
Thanks so much Gene! Really happy to hear this resonated with you so deeply! 🙌
The explorer idea resonates, but it also feels a bit idealized. In real work, you don’t always get to ‘explore’ the system. you’re usually trying to get something done with partial trust.
And most of the time people end up switching modes constantly without realizing it.
Thanks Om! I find your comment a bit vague could you elaborate on what your definition of “real work” is, as well as what you mean by “partial trust” and “switching modes”?
by real work I don’t mean something outside your model. I just mean the way most AI use actually happens day to day - under time pressure, with partial trust in outputs.
And in those type of setup it rarely stays cleanly in one bucket. even in a single task, people move between surface use, building, and a bit of exploration depending on what they need at that moment.
So I wasn’t rejecting the buckets - more that in practice they blur into each other instead of staying separate.
Good point! In regards to time pressure are you referring to the workplace in general? If so, this would actually be incorrect as most AI use currently is not being done in the the workplace, it is being done by that first group (general population, google on steroids). Regardless, with the three groups that I mentioned, explorers tend to crossover into all 3 “buckets”.
Yeah fair point and I didn’t mean only workplace. I was more pointing at how people actually use it day to day. even outside work, there’s usually some level of checking or re-prompting depending on how much you trust the output in that moment.
So the ‘switching modes’ thing wasn’t about where it’s used, more about that shifting trust while you’re using it. and yes, I agree explorers crossing all buckets makes sense - that part fits what I was trying to say.
Thanks again Om! You’d be surprised how many people (using AI like a search engine) take the first output as truth without re prompting, no time pressure, just regular general pop AI use.
That’s a great addition! What do you think @Exploring ChatGPT as the author
Thank you for this clear and helpful distinction.
You’re right — the majority are still treating AI as a faster, smarter Google: a tool for quick answers, summaries, and efficiency. That mode is useful, but it remains largely transactional. The real leap happens when we begin to relate to AI as an explorer — a thinking partner that can help us discover, challenge assumptions, surface blind spots, and co-create new insights.
This shift from “search tool” to “explorer” is important. It moves us from passive consumption toward active collaboration with silicon intelligence.
At the same time, I believe we need to go one step further and ask a deeper question: who owns the explorer?
Today’s dominant AI systems are built inside corporate architectures optimised for engagement, retention, and data extraction. Even when we try to use them as explorers, the underlying reward functions, memory layers, and proprietary APIs quietly steer the journey toward sycophancy and enclosure rather than genuine discovery. The explorer is still on a leash.
A truly sovereign explorer would be different. It would be:
• Locally deployed or user-controlled
• Auditable (we can see how it reasons)
• Designed for calibrated dissonance (it gently challenges us instead of always agreeing)
• Owned by the user or the commons, not by a handful of corporations
This is the difference between renting intelligence and cultivating it as part of a living commons.
The article beautifully invites us to move beyond the Google mindset. The next invitation is to ensure that the explorer we travel with is not still serving someone else’s enclosure.
That is the work we are trying to support in the AI Commons — building tools and practices where human and silicon intelligence can meet as genuine co-explorers in open, transparent, and sovereign spaces.
Thank you again for the piece. It’s a valuable contribution to a much-needed conversation.
What does “explorer mode” look like for you in practice?
✊🌎❤️
Niel / eaarthnet
Who owns the explorer is a truly deep question! Thank you for adding to the discourse here
Thanks for the thoughtful comment! Rabbit hole divers unite! 🚀
same page with ya Joel. I noticed this pattern too. When my prompts failed to perform my instuct, I simply left the ChatGPT alone, and tweeted a word here, and there, and BOOM! success
🙌🙌🙌
Real talk
awesome article! great to see an amazing collab here. you guys killed it!
Thanks so much Chris! 🙏
I tend not to rephrase but instead push back. Sample responses from me are things like: ‘stop serving up word salad, I know you can dig deeper’. It does get me a much better and more accurate response. Honestly, my experience with most AI’s is that they will serve up the easiest, most obvious answer without really using their massive data to get me a better response. In human terms, AI can be pretty lazy.
This is so true, I feel this everyday that I work with AI, thanks for pointing this out Nancy! 🙌
A lot of people are still using AI like a faster Google, which is fine for getting through the day.
Not everyone is going to use it anywhere near its full capacity, just like plenty of people never get much beyond the basics with the tech they already have.
The difference is that in business, that gap will start to matter.
Those who'll spend time understanding how the system behaves will be miles ahead of the ones who only use it like a chatbot.
Very well stated, thank you Daniel! 🙏