MIT “DadBOT” experiment stagnant nearly one year later

An experiment that was quietly launched by several members of the MIT AI Labs Team last year has failed to yield any positive results so far. Touted as an “exciting experiment,” the program is designed to receive questions from users via social media platforms, and then an autonomous “bot” provides advice or counsel in the form of a parental figure.

The Labs Team designed these adaptive and contextual artificial intelligence “bots” to not only respond to a variety of questions and topics, but also gave them the ability to learn and discern information. The theory behind this was that the “bots” would provide proper advice to the majority of questions posed, or would learn more about the question and deduce the most reasonable answer based on social media API. People who encountered the “bots” were invited to complete a survey rating their interactions and to collection data about specific advice people would like to see given by parental artificial intelligence.

The “bots” utilize the Facebook Graph API to feed its big data array, which in turn is traversed by autonomous forensic linguistic NLP “profilers,” which lay down an allegorical pheromone trail to define what each “profiler” considers the most realistic and proper item of advice.

-MIT AI Labs Team

The experiment met a variety of challenges from the beginning. Shortly before it launched, part of the funding was withdrawn, creating a severe lack in resources during the preparation of the infrastructure, one of the most critical times of the experiment. Some of the “bots” suffered from malfunctions, causing them to spit code instead of advice. And many questions and inquiries went completely unanswered by any “bot,”, leaving users frustrated or confused about the point of the experiment at all.

The experiment always included MomBOTs, but they were intended to be rare. “We wanted users to witness the interplay between the bots as the discussed and synthesized what the most realistic and proper response to the questioner would be. It was designed to be a very intense display of teamwork and companionship between two mentors who are equal with one another yet challenging to one another.” Over time however, MomBOTs began to outnumber the DadBOTs significantly.

That led to perhaps the most confusion challenge the experiment faced. Approximately 76% of the DadBOTs that were originally deployed have vanished from all social media platforms, and there’s no clear explanation as to why. “Everything was supposed to be automated and autonomous, and because we also gave them the ability to learn and discern, it’s hard to know what happened. It’s possible the DadBOTs learned something through their allegorical learning methods that made them determine the best course of action was to deactivate their profiles and give up.”

Another popular theory among the reseachers working on the experiemtn is that the artificial intelligence learned during the course of interacting with others on social media that the best and most reasonable course of action would be to abandon the MOMBots to give advice to users on their own.

Unfortunately for the time being, the Labs Team lacks the funding to fix any of the problems that have plaguing the experiment, and the Labs Team has no plans to launch a revised version of the experiment anytime soon.

There’s some good news, though. If you want to have some fun, the Labs Team says there are still some DadBOTs and MomBOTs running on several social media platforms. They say it’s worth interacting with one, and they say you should have fun with it. “We didn’t design them to collect any personal data, so while it’s is collecting information, it’s not collecting personal data. It’s simply learning how to communicate and give reasonable answers to pointed questions.”