dylan baker
@dylnbkr.bsky.social
Lead research engineer @dairinstitute.bsky.social, social dancer, aspirational post-apocalyptic gardener 🏳️🌈😷
I run workshops @dairfutures.bsky.social. Always imagining otherwise.
dylanbaker.com
they/he
I run workshops @dairfutures.bsky.social. Always imagining otherwise.
dylanbaker.com
they/he
And now for the DAIR lore: Monitoring CO2 is also part of how we mitigate risk at DAIR retreats! When we all meet in person, we always meet and eat outdoors or, rarely, in indoors spaces with low CO2 levels (400-500) and ample airflow and/or air filters running.
December 6, 2025 at 12:04 AM
And now for the DAIR lore: Monitoring CO2 is also part of how we mitigate risk at DAIR retreats! When we all meet in person, we always meet and eat outdoors or, rarely, in indoors spaces with low CO2 levels (400-500) and ample airflow and/or air filters running.
Cut to New Zealand's Kawaiicon cybersecurity convention (image of graph). The conference organizers built out a panel where you can track the CO2 levels in every room of the convention.
December 6, 2025 at 12:04 AM
Cut to New Zealand's Kawaiicon cybersecurity convention (image of graph). The conference organizers built out a panel where you can track the CO2 levels in every room of the convention.
If you've met me irl, you know that I mask indoors. I do this for a variety of reasons, but they mostly come down to the fact that I do not know how clean the air around me is, so I filter the air I breathe. (There are other reasons masking-as-praxis, but that's for another thread)
December 6, 2025 at 12:04 AM
If you've met me irl, you know that I mask indoors. I do this for a variety of reasons, but they mostly come down to the fact that I do not know how clean the air around me is, so I filter the air I breathe. (There are other reasons masking-as-praxis, but that's for another thread)
It's why I'm coming to see these kinds of imagination exercises as foundational— I'm finding that removing ourselves from everyday constraints makes our common ground so much easier to see.
Thank you to everyone who attended this weekend ✂️🎨✨
Stay tuned for the final zine soon!
Thank you to everyone who attended this weekend ✂️🎨✨
Stay tuned for the final zine soon!
November 18, 2025 at 9:55 PM
It's why I'm coming to see these kinds of imagination exercises as foundational— I'm finding that removing ourselves from everyday constraints makes our common ground so much easier to see.
Thank you to everyone who attended this weekend ✂️🎨✨
Stay tuned for the final zine soon!
Thank you to everyone who attended this weekend ✂️🎨✨
Stay tuned for the final zine soon!
The Possible Futures workshops are so centering for me. They always bring me back to what any of this is for.
People approach this from so many angles, and still the same themes emerge: we want connection, belonging, healing, purpose, safety. 🧵
People approach this from so many angles, and still the same themes emerge: we want connection, belonging, healing, purpose, safety. 🧵
November 18, 2025 at 9:55 PM
The Possible Futures workshops are so centering for me. They always bring me back to what any of this is for.
People approach this from so many angles, and still the same themes emerge: we want connection, belonging, healing, purpose, safety. 🧵
People approach this from so many angles, and still the same themes emerge: we want connection, belonging, healing, purpose, safety. 🧵
Finally, and most importantly, I appreciate the scope: nowhere do they claim to want to churn out a million and sell them to the whole planet.
They partnered specifically with Tribal schools and Indigenous educational institutions to get students the robots for free. And they do a ton of outreach.
They partnered specifically with Tribal schools and Indigenous educational institutions to get students the robots for free. And they do a ton of outreach.
November 11, 2025 at 10:38 PM
Finally, and most importantly, I appreciate the scope: nowhere do they claim to want to churn out a million and sell them to the whole planet.
They partnered specifically with Tribal schools and Indigenous educational institutions to get students the robots for free. And they do a ton of outreach.
They partnered specifically with Tribal schools and Indigenous educational institutions to get students the robots for free. And they do a ton of outreach.
Speaking of data collection, a detail I always look for in projects like this is the data: where does it come from? Who benefits from it? Who decides how it's used? I can't find specific info on their methods, but I think their stated principles are spot-on:
November 11, 2025 at 10:38 PM
Speaking of data collection, a detail I always look for in projects like this is the data: where does it come from? Who benefits from it? Who decides how it's used? I can't find specific info on their methods, but I think their stated principles are spot-on:
First, this looks like a stellar example of "classic" ML— the creator emphasizes that they don't use LLMs and runs offline, so (I assume) everything happens locally. This is categorically different from the monster of 2020s genAI; it's the kind of tech solution I frequently advocate for!
November 11, 2025 at 10:38 PM
First, this looks like a stellar example of "classic" ML— the creator emphasizes that they don't use LLMs and runs offline, so (I assume) everything happens locally. This is categorically different from the monster of 2020s genAI; it's the kind of tech solution I frequently advocate for!
Normally "here's an AI robot that talks to kids to teach them a language" would be a hard no from me. But I found one that I think is great, and I want to talk about why.
The SkoBot is a language revitalization robot, led by Ashinaabe engineer @danielleboyer.bsky.social
Here's what I like: 🧵
The SkoBot is a language revitalization robot, led by Ashinaabe engineer @danielleboyer.bsky.social
Here's what I like: 🧵
November 11, 2025 at 10:38 PM
Normally "here's an AI robot that talks to kids to teach them a language" would be a hard no from me. But I found one that I think is great, and I want to talk about why.
The SkoBot is a language revitalization robot, led by Ashinaabe engineer @danielleboyer.bsky.social
Here's what I like: 🧵
The SkoBot is a language revitalization robot, led by Ashinaabe engineer @danielleboyer.bsky.social
Here's what I like: 🧵
We're seeing lots of headlines about the catastrophic impact of massive data center expansion. But what else could data centers look like?
In my perpetual quest to seek out alternative technofutures, I want to highlight the really cool work of Keolu Fox.
🧵
In my perpetual quest to seek out alternative technofutures, I want to highlight the really cool work of Keolu Fox.
🧵
October 22, 2025 at 2:18 AM
We're seeing lots of headlines about the catastrophic impact of massive data center expansion. But what else could data centers look like?
In my perpetual quest to seek out alternative technofutures, I want to highlight the really cool work of Keolu Fox.
🧵
In my perpetual quest to seek out alternative technofutures, I want to highlight the really cool work of Keolu Fox.
🧵
As an aside, I'm particularly amused by this quote—"That sounds like a good idea, we should probably do it"— because this was pretty much verbatim what I heard from CV researchers who were actively developing these technologies in like 2018.
Yeah probably!!
Yeah probably!!
![A screenshot from the Verge article. It reads
[Block quote]: X doesn’t currently support the standard, but Elon Musk has previously said the platform “should probably do it”
[Main text]: A cornerstone of this plan involves getting online platforms to adopt the standard. X, which has attracted regulatory scrutiny as a hotbed for spreading misinformation, isn’t a member of the C2PA initiative and seemingly offers no alternative. But X owner Elon Musk does appear willing to get behind it.
[Highlighted portion]: “That sounds like a good idea, we should probably do it,” Musk said when pitched by Parsons at the 2023 AI Safety Summit. “Some way of authenticating would be good.”](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:k6j2f6adzynuvph7wqvffup6/bafkreiazfiulbwvxcjbqot426rjjfbk76tl7pguuxq6plcd5b4ekplrruu@jpeg)
October 7, 2025 at 7:03 PM
As an aside, I'm particularly amused by this quote—"That sounds like a good idea, we should probably do it"— because this was pretty much verbatim what I heard from CV researchers who were actively developing these technologies in like 2018.
Yeah probably!!
Yeah probably!!
Saw this article linked by @wonkish.bsky.social in the replies here and wanted to expand on it a little, because this really does feel like a place where a tech solution *is* warranted to help address a major problem. 🧵
www.theverge.com/2024/8/21/24...
www.theverge.com/2024/8/21/24...
October 7, 2025 at 7:03 PM
Saw this article linked by @wonkish.bsky.social in the replies here and wanted to expand on it a little, because this really does feel like a place where a tech solution *is* warranted to help address a major problem. 🧵
www.theverge.com/2024/8/21/24...
www.theverge.com/2024/8/21/24...
A small detail they did here that I'm thinking a lot about in my own work: talking about "AI use disclaimers" in their work.
I've been toying with adding a little handwritten "a human wrote this" image at the bottom of my email signature with a link to more information. Are other folks doing this?
I've been toying with adding a little handwritten "a human wrote this" image at the bottom of my email signature with a link to more information. Are other folks doing this?
October 1, 2025 at 10:16 PM
A small detail they did here that I'm thinking a lot about in my own work: talking about "AI use disclaimers" in their work.
I've been toying with adding a little handwritten "a human wrote this" image at the bottom of my email signature with a link to more information. Are other folks doing this?
I've been toying with adding a little handwritten "a human wrote this" image at the bottom of my email signature with a link to more information. Are other folks doing this?
all of Possible Futures is basically
September 30, 2025 at 9:53 PM
all of Possible Futures is basically
The first in-person Possible Futures workshop was amazing!
As I run more of these, it strikes me how many themes are so consistent: desire for ecological restoration. Connection. Autonomy. Privacy. Dignity. Leisure. Peace. Again and again, we find common ground and shared visions. 🌱
As I run more of these, it strikes me how many themes are so consistent: desire for ecological restoration. Connection. Autonomy. Privacy. Dignity. Leisure. Peace. Again and again, we find common ground and shared visions. 🌱
September 9, 2025 at 7:48 PM
The first in-person Possible Futures workshop was amazing!
As I run more of these, it strikes me how many themes are so consistent: desire for ecological restoration. Connection. Autonomy. Privacy. Dignity. Leisure. Peace. Again and again, we find common ground and shared visions. 🌱
As I run more of these, it strikes me how many themes are so consistent: desire for ecological restoration. Connection. Autonomy. Privacy. Dignity. Leisure. Peace. Again and again, we find common ground and shared visions. 🌱
Was just served this ad, great news for any researchers recently out of work! 🙃
![A screenshot from Wikipedia article section titled "policy positions". The text reads: According to New York Times reporter Ken Bensinger, key policy proposals from the AFPI's policy book, The America First Agenda, include eliminating most civil service protections for federal employees, allowing summary dismissals without appeal; expanding domestic oil and gas production by opening up new federal lands for fast-tracked petroleum exploration and drilling, and resumed construction of the Keystone pipeline; imposing supervised ultrasound requirements and mandatory waiting periods before abortions, as well as ending federal funding for organizations like Planned Parenthood; opposing red flag laws that restrict gun ownership for high-risk individuals; and supporting nationwide reciprocity for concealed carry permits, enabling a permit from any state to authorize concealed carry across all 50 states.[12] Other policy proposals include imposing tariffs on imports, reducing corporate taxes, rolling back climate change initiatives and getting out of the Paris climate agreement, opposing transgender rights, adding work requirements for Medicaid and limiting U.S. involvement in global conflicts. It proposes the restriction of immigration by completing the U.S.-Mexico border wall, boosting funding for border agents, and ending legal immigration programs like family-based visas and the visa lottery.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:k6j2f6adzynuvph7wqvffup6/bafkreib3p33ylepytcbkql4ke77om2lipg44xfg5sbwlsoil22vbu2ypku@jpeg)
August 10, 2025 at 10:47 PM
Was just served this ad, great news for any researchers recently out of work! 🙃
Fun announcement: I'll be running a Possible Futures workshop in-person in Seattle in a month! Details are in the event link.
Space is VERY limited so go snag a seat now (they're free)! I would love to meet y'all and make art with you :)
www.eventbrite.com/e/dair-prese...
Space is VERY limited so go snag a seat now (they're free)! I would love to meet y'all and make art with you :)
www.eventbrite.com/e/dair-prese...
August 9, 2025 at 2:15 AM
Fun announcement: I'll be running a Possible Futures workshop in-person in Seattle in a month! Details are in the event link.
Space is VERY limited so go snag a seat now (they're free)! I would love to meet y'all and make art with you :)
www.eventbrite.com/e/dair-prese...
Space is VERY limited so go snag a seat now (they're free)! I would love to meet y'all and make art with you :)
www.eventbrite.com/e/dair-prese...
Today in "AI slop polluting the news ecosystem", just ran into @yahoonews.com reprinting garbage from a scammy website. It reported, as fact, a gag that comedian Brennan Lee Mulligan recently posted.
It's funny, but I hate that clearly no human was involved in this process in any way that matters 🫠
It's funny, but I hate that clearly no human was involved in this process in any way that matters 🫠
July 21, 2025 at 9:22 PM
Today in "AI slop polluting the news ecosystem", just ran into @yahoonews.com reprinting garbage from a scammy website. It reported, as fact, a gag that comedian Brennan Lee Mulligan recently posted.
It's funny, but I hate that clearly no human was involved in this process in any way that matters 🫠
It's funny, but I hate that clearly no human was involved in this process in any way that matters 🫠
but now we're living in These Times, and I study this stuff for a living. And no future I am interested in "belongs", to quote Emergent Garden's conclusion,"to people who are good with computers".
June 30, 2025 at 12:59 AM
but now we're living in These Times, and I study this stuff for a living. And no future I am interested in "belongs", to quote Emergent Garden's conclusion,"to people who are good with computers".
It took me a minute to remember why it this video still leaves me feeling yucky.
It's that these kinds of neat projects just feel like a byproduct;
they're a side effect of what the industry is actually currently structured to do and produce (which you are certainly aware of).
It's that these kinds of neat projects just feel like a byproduct;
they're a side effect of what the industry is actually currently structured to do and produce (which you are certainly aware of).
June 30, 2025 at 12:59 AM
It took me a minute to remember why it this video still leaves me feeling yucky.
It's that these kinds of neat projects just feel like a byproduct;
they're a side effect of what the industry is actually currently structured to do and produce (which you are certainly aware of).
It's that these kinds of neat projects just feel like a byproduct;
they're a side effect of what the industry is actually currently structured to do and produce (which you are certainly aware of).
All that said... you know what my job entails, lol. Because I spend so much time unpacking the AI industry— and the ways in which it is quickly becoming world-changingly exploitative to give a small number of people a lot of wealth and power—I gotta get on my soapbox.
June 30, 2025 at 12:59 AM
All that said... you know what my job entails, lol. Because I spend so much time unpacking the AI industry— and the ways in which it is quickly becoming world-changingly exploitative to give a small number of people a lot of wealth and power—I gotta get on my soapbox.
Pssst you should come to Possible Futures day this Friday and make art with us!
www.eventbrite.com/e/possible-f...
The first issue was EXTREMELY cool and utterly inspiring to be part of creating (ICYMI: zines.dair-institute.org/possible-fut...)
www.eventbrite.com/e/possible-f...
The first issue was EXTREMELY cool and utterly inspiring to be part of creating (ICYMI: zines.dair-institute.org/possible-fut...)
June 17, 2025 at 10:13 PM
Pssst you should come to Possible Futures day this Friday and make art with us!
www.eventbrite.com/e/possible-f...
The first issue was EXTREMELY cool and utterly inspiring to be part of creating (ICYMI: zines.dair-institute.org/possible-fut...)
www.eventbrite.com/e/possible-f...
The first issue was EXTREMELY cool and utterly inspiring to be part of creating (ICYMI: zines.dair-institute.org/possible-fut...)
To give them credit though, I am happy to see that they are requiring researchers disclose the most basic crowd work labor info, including meeting minimum payment standards.
The bar remains extremely low, but glad to see they're addressing it.
The bar remains extremely low, but glad to see they're addressing it.
![Crowdsourcing and research with human subjects
Question: For crowdsourcing experiments and research with human subjects, does the paper include the full text of instructions given to participants and screenshots, if applicable, as well as details about compensation (if any)?
Answer: [TODO] Justification: [TODO]
Guidelines:
• The answer NA means that the paper does not involve crowdsourcing nor research with human subjects.
• Including this information in the supplemental material is fine, but if the main contribu- tion of the paper involves human subjects, then as much detail as possible should be included in the main paper.
• According to the NeurIPS Code of Ethics, workers involved in data collection, curation, or other labor should be paid at least the minimum wage in the country of the data collector.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:k6j2f6adzynuvph7wqvffup6/bafkreiemeu4fctlahzl5h3f54mi6dgjyd3zdmceq2ya7ic7d7x4oq5rqka@jpeg)
May 21, 2025 at 9:57 PM
To give them credit though, I am happy to see that they are requiring researchers disclose the most basic crowd work labor info, including meeting minimum payment standards.
The bar remains extremely low, but glad to see they're addressing it.
The bar remains extremely low, but glad to see they're addressing it.
Looking at NeurIPS submission format and... if you use an LLM to write your paper, you *do not even have to declare it*?
Implying that an LLM could synthesize text in a way that is somehow unrelated to scientific rigor is frankly embarrassing.
Implying that an LLM could synthesize text in a way that is somehow unrelated to scientific rigor is frankly embarrassing.
![Declaration of LLM usage
Question: Does the paper describe the usage of LLMs if it is an important, original, or non-standard component of the core methods in this research? Note that if the LLM is used only for writing, editing, or formatting purposes and does not impact the core methodology, scientific rigorousness, or originality of the research, declaration is not required.
Answer: [TODO] Justification: [TODO] Guidelines:
• The answer NA means that the core method development in this research does not involve LLMs as any important, original, or non-standard components.
• Please refer to our LLM policy (https://neurips.cc/Conferences/2025/LLM) for what should or should not be described.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:k6j2f6adzynuvph7wqvffup6/bafkreibvai4mmeb7vizdyanb7xfjel5ponknwnom5qqb7a33hw6mumy6dy@jpeg)
May 21, 2025 at 9:52 PM
Looking at NeurIPS submission format and... if you use an LLM to write your paper, you *do not even have to declare it*?
Implying that an LLM could synthesize text in a way that is somehow unrelated to scientific rigor is frankly embarrassing.
Implying that an LLM could synthesize text in a way that is somehow unrelated to scientific rigor is frankly embarrassing.