brochure : raw version
Join the discord: https://discord.gg/F6g9sKGst Calendar Events: [click to subscribe to google calendar] Video series: https://course.groundless.ai
Whatever you may have come across about this event so far may have sounded complicated, but the task for you is pretty straightforward:
That's all, really. Everything else is geared toward helping you make these two submissions good.
In terms of executing both submissions above, we'll give you some practical guidance, and then some more subtle perspectives.
(We hope that at least some of you will see beyond the coding and the prizes and connect with this deep questioning of what "AI" even is, how perverse the risks may become with astonishing-yet-jagged intelligence, how you and your attention matter especially in this era, and why "solving" or systematizing may remain insufficient and even exacerbate issues.)
Regardless, let's get you rolling already with logistics you can do right away:
Now let's dive into details.
For the two tasks A and B mentioned above:
3 (recorded) meetings.
To keep things straight, we will refer to you as giver, and the friend that you're going to work with as receiver. (You might also have assistant givers forming a team, but we'll assume it's just you for ease of explanation.)
You'll schedule 3 meetings with the receiver. You'll record them by inviting the official Fireflies account ([email protected]) to the call.
During the 3 meetings, the giver (you) is generally angling towards taking ownership of an issue that plagues the receiver as they go about their day on their device.
For example, it might be the specific way that the receiver scrolls around on a google doc when writing drafts.
In each meeting the giver is deeply accompanying the receiver, which can involve collaboration between giver and receiver on i) discovery of what is relevant/irrelevant to the receiver in their digital environment and ii) design of a habit, structure, or interface that fits an issue of the receiver. There is no prescribed length for the meetings but expect to spend roughly 30-90 minutes per session.
In the example above, perhaps you both discover that the receiver tends to get excited about ideas while writing in the doc and tries to cram them in the draft. You may end up building a clean little "ideas drawer" that allows the receiver to jot ideas into and return to the doc, with the receiver's aesthetic in mind.
Both the discovery and design are meant to be real and helpful and useful over time for the receiver, not hypothetical or hacky or (self-)deceptive. This is why we have three sessions, so there can be extended dialogue and testing for proper integration. (There will be separate time windows to submit each session recording. That ensures they are sufficiently spread apart rather than being crammed into one day near the end.)
Again, in the example above, perhaps by the second session you both already notice that the "ideas drawer" is quite addictive and ironically causing more distraction. You have a real conversation with the receiver about their intentions, and meet some general avoidance patterns in writing. You propose a simple pomodoro for attention regulation instead, which google docs already has, and also limit the ideas drawer per session. In the third session it is clear that the ideas drawer wasn't helpful and should be sadly retired, but creating an elegant version of the pomodoro that auto-starts would do a lot.
As may be obvious from the example, we are less interested in flashy overpowered one-and-done gimmicks and more in simple things that anyone could have built (but, importantly, didn't). While interesting ideas and aesthetics and functionality may play a role, it is important to remember that the real judge here is, in a sense, whether the user is set up to have real and lasting benefits by the end rather than left worse off. How we'll evaluate is discussed in the Prizes and Evaluation section below.
This whole exercise is mainly a way to bring attentiveness to an issue, rather than overload the receiver with more things to do and less energy. (Of course, this does not preclude that in many cases it may be wise for the receiver to overhaul how they do things in order to reduce dependence, including activities outside of their device.) You may listen to the receiver talk about their issue, but it is best to drop into their day and observe them directly and notice any mindless switching, scrolling etc. You are encouraged to explore your own style once you've been through the video series and understand what the focus is.
Since these sessions are recorded, don't forget to inform the receiver as the very first step. (They won't be shared outside the evaluation process unless you both choose so.) Other disclaimers you should make are in the FAQ section below.
1 post.
Short Version
The structure and content of your post output is aimed neither at casual humans nor machines, but another giver-receiver pair to make use of.
The result is that while you get to create "soloware" for just one person, you also share sketches of it clearly so that many others who have slightly different issues may still "steal it from a screenshot" and remix and recreate it for their specific use case.
For example, you may have shared your journey of the "ideas drawer" (including how it was later replaced with an auto-start pomodoro). Another giver-receiver pair utilizes the same concept, but for email drafting for someone who writes a lot of email. For this receiver, the limit and pomodoro may be irrelevant, since it is just email rather than creative writing. It is matched to their aesthetic and workflow over 3 sessions.
Make sure to include whatever you can in your post, in whatever format, that can help some other giver-receiver pair steal your ideas, code, insights, wires, sketches, screenshots, habits etc. To help with this, we have some guidelines for you (the giver) in the Prizes and Evaluation section below.
Long Version
This task is titled "1 post" but it's not really about writing a standard assigned report to prove that you've done what was asked. It's actually intended to be a new kind of artefact in a new kind of infrastructure. Ultimately, it may not feel too different from the activity of writing about a project you've been working on, but it may be helpful to understand a little bit about what's behind this, which you'll see below.
After 3 sessions you may have substantially or at least slightly helped the receiver. The benefit may have come from being witness to frenzied interaction, loosening and questioning sticky frames, redirecting and freeing the receiver's attentiveness, (re)structuring their flow, or even just discovering something together. It may have involved building something.
It is time now to distribute these benefits. To scale them for many people. This may sound bizarre, since everything so far has been about a one-to-one interaction that is about a highly specific issue with a single receiver and their idiosyncrasies.
One of the core principles at Groundless Alignment (the organization behind this project) is generalization via specificity. "General" and "specific" are usually treated as opposites; if something is designed to be highly specific for one person then it is assumed that it can not scale or generalize to many. To do that you will need to templatize or parametrize the specific instance within some general system. For example, the prerequisite for distribution of an app is consolidation into i) a generalized abstraction that fits an entire class of people, a userbase, rather than only a single specific user and ii) a standardized package fitted to the distribution ecosystem, such as an app store.
These two things (abstraction and distribution) depend on each other. If many people couldn't easily download and install an application package, there wouldn't be a point in abstracting it for many people. And vice versa, if there wasn't an abstraction that worked for many people, few would bother cohering and uploading a package, let alone downloading and installing one.
We will upgrade both i) and ii) simultaneously to look into the future of interfaces, interface stores, and open source culture under ambient intelligence. Cheap and fast intelligent infrastructure is often hard to think about — typically we will just imagine a chatbot or out-of-sight-out-of-mind automation — but we will make use of the "generalization via specificity" lens to go in a very different direction.
What does it look like, to generalize via specificity for apps and app stores? Instead of standardized infra for storing, downloading and installing a package locally, we can actually recreate software locally from the ground up in increasingly little time with no fixed protocols. With AI-enabled recreation, we can make use of screenshots and descriptions and what is valuable, rather than strictly formal code. Most importantly, since it does not need to be the same package when recreated, instead of abstracting patterns for a mass of users, we can merely make something highly specific that can inspire and seed a customized recreation for someone else. We can scale things without making them formulaic or cookie-cutter.
The result is that while you get to create "soloware" for just one person, you also share sketches of it clearly so that many others who have slightly different issues may still "steal it from a screenshot" and remix and recreate it for their specific use case.
For example, you may have shared your journey of the "ideas drawer" (including how it was later replaced with an auto-start pomodoro). Another giver-receiver pair utilizes the same concept, but for email drafting for someone who writes a lot of email. For this receiver, the limit may be functional, since it is just email. It is matched to their aesthetic over 3 sessions.
What's happening here is we scale specificity directly rather than first abstracting it for an imagined audience. Our fears of what AI does to design is turned 180. Where "slop" is almost by definition generic and shallow, this a move towards the opposite: specificity and subtlety, more than even normal software. Additionally, human taste and interaction is central here, in both creation and curation. It's a mere quantitative change in a way, but a completely different world. Kind of like "the internet' is just a very fast and cheap telegraph, yet totally different.
While that may be exciting, the relevance of human interaction shows up in a deeper way. What we build is secondary to waking up humans from fantasies of automating away their lives and meaning, as you'll find the more you engage with this project.
That's all very speculative, and this section is meant to be practical. Luckily this is quite simple practically: your post is this "stealable" artefact. So make sure to include whatever you can in your post, in whatever format, that can help some other giver-receiver pair steal your ideas, code, insights, wires, sketches, screenshots, habits etc. To help with this, we have some guidelines for you (the giver) in the Prizes and Evaluation section below. Eventually, we want this to be a legitimate economy so that "stealing" and remixing is sustainable and rewarding, which is why we may support you to post your artefact as part of your own (paid) substack :)
Video series and other material (~1 hour)
There is a set of videos that you must complete watching before you begin. You can find them here: course.groundless.ai.
Note, none of this content, certainly not the videos, is well suited for AI-summarization.
Q. How is this different from a regular hackathon? Why not just do a regular hackathon?
Hackathons even pre-AI tend to reward novelty and flashiness over something that will actually integrate. In some ways, we don't even care about what you build, as much as helping your friends pay attention to what matters to them in harsh environments where they are assaulted by notifications and anxious work demands, for example.
Q. How is this related to AI safety?
You'll see more later but in short: our threat model is that everything that is happening around AI is shaped by our "frames" (perspectives and motives, in each moment). These moment-to-moment frames are more invisible but more relevant than systems. For each person, a big and ever-present risk is codependent relationships and subtle frame control from AI systems as they get smarter. The contact point between a human and a system, where all this action and predation may happen, is what an "interface" is. If this practice of alertness to frames and what really matters grows in a robust way, it may accord frame-literacy and free the minds that shape the systems, policies, visions and even dangers to come.
Q. How many hours is this? What is the schedule like, if it's not like a typical weekend hackathon?
The actual hours you put in will vary a lot based on what you end up doing and how you'd like to show up.
Here's an hours breakdown, which will be spread out over 3 weeks:
| Activity | Hours |
|---|---|
| Perusing video series and other material | 2 hours |
| The main (recorded) sessions with receiver | 1-4 hours |
| Tweaking any design or structure by yourself | 0-10 hours |
| Writing the post | 2-10 hours |
| Optional events | 0-6 hours |
| Total | 5-36 hours |
See the Schedule section below for the full schedule.
Q, What do I get if I do this well?
Some of you might really get what this is about and why this is important. Done well, many people may benefit from a single good giver over time. We intend to support those who show promise to do that sustainably, for a living.
We also have $5,000 worth of prizes for this event. You can see more in the Prizes and Evaluation section below.
Q. What prerequisites do I need to be a participant?
Primarily, sincerity, openness of mind, and commitment for the time needed (stated above). Sincerity is helpful to remain an alive human who can listen keenly, in the sense of both i) comprehension of the subtleties of the content here and ii) attunement to the receiver you will be helping. It is also extremely relevant to being risk-aware. Openness of mind is helpful to see past the assumptions and perspectives you may have locked into already around "AI" or "AI safety". And of course, commitment is critical for any of this to work out, since the process needs to mature over time and the details are king.
If you have a penchant for depth of listening, the kind that wishes well and asks for no change, or general skills in relating to or coaching people without treating them as objects or algorithms to be optimized, you are probably a good fit.
It may help to have some comfort with technical language or software design or development, but it isn't strictly necessary. In fact, if you have a lot of technical manipulation chops but not much ability to be with a person without "fixing" them, you will likely exacerbate the problem on our threat model.
Q. Am I supposed to organize these different activities myself? And what are the activities again?
There is a schedule that should orient you, but also a lot of freedom over the three weeks. The main activities, to recall, are:
The main thing to be oriented to is the submission windows for the recordings of the three meeings (Sep 6th, Sep 11th EoD AoE, and Sep 16th EoD AoE). They will space out your meetings with the receiver. After you're done with those, you will have several days to write your post and then submit it by Sep 21st EoD AoE.
Q. What do I do if I have questions about this?
Ask on the Discord channel.
Q. What is a "receiver"?
The "receiver" is a person you will invite to have three calls with you and receive your support.
Q. What sort of person is best to be a receiver?
The minimal requirement is just that they commit to three meetings with you and share some aspect of their digital life. Ideally they are someone who is
Q. What should I be disclaiming to my receiver?
It is impossible to guarantee real help, but try your best to avoid making things worse!
Q. How do I find a receiver?
Anyone who fits the bill will do. You could ask a friend or family member, or even an acquaintance online. But make sure they understand the above and can commit to 3 meetings with you. Best to reach out as quickly as possible, since there will be very specific time windows (see the Schedule section below) to submit the call recordings!
Q. Does my receiver have to sign up in some way or read and understand all the details here?
No, they don't have to do either. However, they might benefit quite a bit from taking a look at this document, or may just be curious. Feel free to send it to them, but the giver (you) are generally taking responsibility to welcome them into this experience.
Q. What's in it for a receiver, again?
Most people who use AI (usually in chatbot form) are actually forced to be ad hoc designers. They have to command a rather general system and will tend to do it sort of cluelessly or mindlessly or whichever way is least painful. They fail to be considerate to themselves and their holistic benefit. They are implicit designers, just poor ones, and in a hurry.
Separately, nearly everyone is in overwhelm by default in their digital environments. When we describe the issue of "some people click into their email app with the wish to send someone an email, then find an important email and start replying to it, and emerge later having forgotten why they opened it all", pretty much 100% of people relate. Moreover, they almost always think they're the only one with this issue!
The two points above combine to make it quite compelling to be given company in all the neurotic digital messes we find ourselves in. People who can benefit from such company (i.e. the receivers who can benefit from a giver's company) may not yet be alert to the coming increasingly personalized and intelligent predation that will make this all the more challenging, but this will be a perfect opportunity to develop that alertness.
Q. What do I say to prospective receivers? How do I invite them in?
You could share the answer to the last question with someone you know and see if they would like that kind of company as an experiment. With someone with more space or energy, you could send them a link to this whole document and ask them about being on the receiving end.
You could keep it simple and just ask "Would you like to do a few calls with me about what might help you in your day-to-day computer/phone life?" but make sure to give them the background and disclaimers!
Q. Won't friends or family be biased as receivers?
brochure : AI assisted version