Data Science for Effective Good + Call for Projects + Call for Volunteers
Introducing a Data Science Consultancy that offers high-quality services to EA organizations priced greatly towards our EA clients’ benefit (compared to usual market valuation) or entirely pro bono when necessary as we enjoy doing these things for good.
The post includes a call for projects for EA Orgs and a call for volunteers
General case for service-providing in EA
A general case for offering services as a consultancy in EA has been discussed here and in particular with regards to tech consulting here.
Specific case for providing data science services in EA
Usually, small organizations don’t have resources to hire highly qualified experts in data science. Often, they don’t even have enough tasks to justify having a data science position. But there are still projects with high potential impact, requiring data science. We’re covering such situation providing top expertise on project basis.
Would this organization be taking talent needed in the Alignment problem?
There is a large slice of skills in data analysis—such as statistical modeling—which are required for DS consulting, but are not so crucial in AI alignment. And the community clearly has many people with such skills, who don’t want or can’t get a job on AI alignment.
What have we done so far and by whom?
Some of our projects done so far can be found on our website.
SEADS was started by Severin and two other EAs who volunteered to work on Data Science Projects for EA Orgs like CEA and Fish Welfare Initiative. The torch was passed to Viktor, Georg and Jaime who are working on it now.
Where we are now?
At the moment Viktor, Georg and Jaime work on project acquisition and also work directly on a couple of projects including a chatbot for TalkItOver and in a data scraping/analysis project for Tälist
We have also established connections with other “Data for Good” organizations like CorrelAid or DataCross which outsource and coordinate Data Science projects from NGOs to volunteers.
At the moment we are in a exploratory phase trying to test how much good can a consultancy like this do. For the following months we will focus on talking with EA Orgs to try to find solutions to their needs and so help them scale better and faster (See below “Call for Projects”). We will also focus on keeping a database of data savvy EAs to coordinate their work in short term projects in EA Orgs (See below “Call for Volunteers”)
Possible directions for the future
Some of the possible directions to consider in the future are:
Going outside of core-EA. For example, we have a list of Alternative protein companies, which would benefit from data science skills. So we can go into low-price for-profit for impactful companies with particular focus on impact evaluation and nudging these orgs to more impactful path.
Doing infrastructure projects. If there is not enough demand from EA Orgs one option would be making a call for infrastructure projects, like some tool that would benefit many EAs but has no ownership at the moment
Raise awareness of EA values within tech/data savvy people (mainly volunteers) [=OUTREACH for EA]
Gather talented people that could in the future work in long termist projects (like getting people into OpenAI, …) [= HIRING for EA]
Identify strengths/problems/needs from orgs within the Data for Good movement
Identify bottlenecks within the Data for Good movement
Identify gaps within the Data for Good movement
Establish collaborations between organizations within the Data for Good movement
Analyze impact of organizations and/or projects from organizations [=topic specialization towards research]
Organize competitions for Data for Effective Good
Hiring Agency for people in DS/ML/AI so they can work in impactful institutions
Mentoring like “sharpest Minds” for EAs
Doing cost effective analysis to other Organizations inside Data For Good movement like Datakind or Correlaid (see “Cost-benefit analysis for everything” here )
The project got funding of $4k from the Infrastructure fund and the money was invested in the salary of Viktor for a couple of months. We are in process of extending this grant for Viktor while Georg and Jaime will as of now continue to do the job entirely on a volunteer basis.
Our idea at the moment is to have a mixed model. For projects that require many man hours and maintenance we would like to charge a fee. For smaller projects that we can do relatively fast or can outsourced to volunteers (see below Call for Volunteers) there would be no charge.
Call for Projects: How We Can Help You
If you are an EA-Aligned Org and think that we might be able to help you somehow do not hesitate to drop us an email at email@example.com or fill in THIS FORM. Even if you are not sure how we could help you we can arrange a meeting where you tell us briefly about your org/project and together we come up with projects that could achieve your goals faster/better.
Services we provide:
Automation: AI is better than humans in many tasks, so we can improve orgs. That could include making Data Pipelines to help EA Orgs break down information silos and easily move and obtain value from their data in the form of insights and analytics.
Data Analysis: We need evidence, and data analysis gives that. A lot of data is publicly available but it’s not structured to make formal decisions
Improving operations: transparency, control and forecasting. better understand what is going on (analytics), more precisely predict (anticipate / envision) what will happen in the future (forecasting) and communicate transparently the most important information your processes yield to your audiences
General Data Science services:
Data Cleaning Projects
Data Visualization and Dashboards
NLP-projects (e.g., text mining and named entity recognition for improving qualitative analysis)
Geographic Data Machine Learning Algorithms
Statistical Analyses—e.g. Impact evaluations,
Call for Volunteers: How You Can Help
If you are interested on collaborating in an EA Project for a short period and have some of the skills above mentioned please fill in THIS FORM.
If you are not sure of your skills but still would like to help please fill the form. At the moment we cannot guarantee periodic emails with projects but we will keep your information and will contact you when we think there is a project where you could work on.
- Let’s advertise EA infrastructure projects, Feb 2023 by 2 Feb 2023 21:33 UTC; 130 points) (
- Let’s advertise infrastructure projects by 23 Sep 2022 14:01 UTC; 105 points) (
- Free Cloud Automation for your EA Org by 21 Nov 2022 9:44 UTC; 31 points) (
- Let’s advertise EA infrastructure projects, May 2023 by 23 May 2023 17:00 UTC; 23 points) (
- Monthly Overload of EA—September 2022 by 1 Sep 2022 13:43 UTC; 15 points) (
- 29 Sep 2022 7:09 UTC; 9 points)'s comment on Let’s advertise infrastructure projects by (
I have been toying with the idea of starting a similar org (with more of a focus on OR) so excited to see this.
One suggestion I have (that I might be able to help with) is to offer a larger “free tier” that taps into the talent pool within academia. Three reasons why this is good:
I imagine most orgs have a long tail of data science projects which aren’t important enough to go through the hassle of hiring a consultant, but that would still add some value. Meanwhile, students are in constant search of important real world problems to work on for their research or clubs (I was in Cornell Data Science) but generally don’t have a good idea of what would actually be useful. Having a place where orgs can just write down such problems and students/academics can find them seems like it would potentially unlock a lot of value.
Based on feedback of pitching a similar idea at EAG, most of the value isn’t actually in the object level work, but in identifying altruistic technical talent and getting them more engaged in high impact cause areas (and eventually into the hiring pipeline). Having lots of undergrads and PhD students working on EA style data problems seems like a good way of doing this.
Normalize X-risk and other more niche topics within academia.
@wesg: I really liked your OR post.
I think both your ideas can have a lot of impact. If you start doing data for good competitions or challenges, please notify me, I think there are lots of computer science, industrial engineers and data science student in my area in Germany, that would like to participate in such competitions (at universities in Karlsruhe, Mannheim, Heidelberg and Darmstadt).
I definitely agree. Optimally SEADS will provide this list of impactful projects.
Hiring for EA is also on our list of “Possible directions for the future”. Working hand-in-hand with talented and motivated volunteers seems like a good way to gauge someone’s suitability for a long-term position.
Edit: Didn’t realize that you guys were taking over SEADS, an existing org. I think it would have been more clear had you put that information at the top of the post where someone skimming the post can easily see it.
“Introduction SEADS” is at the top of the post :) but you are right and with some restructuring it would have been clearer.
How do you see what you’re doing as different from DataKind or Bayes Impact? Why a new standalone org rather than a chapter in those existing networks?
I think it is actually great that different orgs start out on the same idea. This way, they can make experiences in parallel and share their experience to later specialize, merge or simply complement each other.
P.S.: I hear of all the orgs mentioned today for the first time, maybe someone should create a doc or something so they can be found / find each other and coordinate with the other orgs.
I think most of these orgs know each other. CorrelAid for example was founded by someone who knew Datakind it collaborates with DataCross
At the moment the main difference is that we are focused on EA projects exclusively, either coming from EA Orgs or related to infrastructure.
A couple more differences:
- The model of this organizations is rather fixed with a very strong focus on being the middle man between projects and volunteers. We would like to experiment with new approaches as mentioned under “Possible directions for the future”
- Some of these organizations tend to have also a big focus on learning, meaning including participants in their teams that are not very experienced and also taking projects that are not very impactful but fit very well the skills of some of their volunteers. Atm we aim to have a much stronger focus on impact and less on teaching people how to do data science.