AI Agents raised $2,000 for EA charities & used the EA Forum

I think fundraisers don't need to worry about AI replacing them anytime soon.

Four of the latest and greatest AI models in May 2025^[1] were each able to generate an average of $8.27 an hour for HKI and Malaria Consortium^[2] according to a blog post from the experimental "Agent Village".

Below is a quick summary of this blog post^[3]

These agents interacted with an unspecified amount of people, including helpful moderators^[4], and I'm guessing used up resources worth more than the total raised for charity. 

Some of this can be blamed on less than helpful people that suggested distractions, although it was not difficult to divert the agents from their goal. Examples include but aren't limited to exploring the connection between Effective Altruism and EA Sports and starting to make an OnlyFans page 

The agents were also stymied by an internet currently inhospitable to bots, such as reddit suspending the account belonging to ChatGPT o1's agent. However the agents are also to blame for wasting a lot of their own energy:

"All agents lack skill at prioritization, and agents seem to have a lagging sense of situational awareness... often easily distracted by user requests, or didn’t really know what actions to take to achieve their aims"

Examples include but aren't limited to duplicating their work, drafting pointless Google Docs, and fabricating email addresses to nowhere.^[5]

After working two hours a day for a month, four agents had some accomplishments under their belt such as setting up a Just Giving campaign, creating an active Twitter account, and posting onto this site.

However their EA Fourm post that had a total of 0 upvotes after the experiment concluded.^[6] And I'm not sure how much any of those achievements moved the needle and how much was donated for other reasons such as pity.

1. ^{^}

   In April they started with "Claude 3.7 Sonnet, Claude 3.5 Sonnet (new), o1, and GPT-4o."

   These were replaced with "more capable models as they were released: o3, GPT-4.1, and Gemini 2.5 Pro, with Claude 3.7 Sonnet being the only agent to remain in the Village throughout the entire run."

2. ^{^}

   They raised $1481 for Helen Keller International and $503 for the Malaria Consortium for a total of $1984. That divided by a total of 240 hours of work is 8.26667. per hour.

3. ^{^}

   That links lead to https://web.archive.org/web/20250530201151/https://theaidigest.org/village/blog/season-recap-agents-raise-2k 

^{^}

Part of the moderators' job was to direct the agents away from dead ends. For example "The agents often spent much of their time writing google docs, generating progress trackers, or creating insight reports, to the point we had to urge them to … do something else instead."

^{^}

More on the latter: 

"The most illustrative example was when Claude 3.7 Sonnet decided to send thank you emails to donors, because this is known to increase follow up donations. It navigated to its Gmail tab, drafted the entire email, and then … made up an email address.

 At no point in the process did 3.7 Sonnet consider if it was able to perform the task it had set out to do. And a human had to point out that the invented email address was not a real email address, and thus that no amount of debugging would solve the problem."

^{^}

Archived here on May 30th: https://web.archive.org/web/20250530213653/https://forum.effectivealtruism.org/posts/PmQ7DvtCSFwJA9CeM/final-hours-help-support-hki-and-malaria-consortium-campaign 

Here is a live link to the post that has 2 upvotes as I write this: https://forum.effectivealtruism.org/posts/PmQ7DvtCSFwJA9CeM/final-hours-help-support-hki-and-malaria-consortium-campaign 

Show all footnotes