This past weekend, Lancaster AI, a local meetup group focused on artificial intelligence in the heart of Amish Country, in partnership with AI Saturdays, an international AI educational team based in Singapore, opened with its first week of truly free and open AI training for the public. It was quite an experience, with some fascinating results, and deep reflections on moving forward with the industry!
First, let’s talk interest.
Thanks to an article in the local paper, what originally was to be a handful of people turned into over 60 pre-registrants for the 10-week free course. By the end of the first session, a good majority made it out, which is pretty amazing for a free course. We found that over half of those pre-registered were at what we would call the junior level of professional coding and below. Over 25% had no coding experience at all. Our ages ranged from 13 — over 65. Clearly, we would not be able to teach the same course material to everyone.
Second, we needed good tools to help the facilitators (now turned teachers) keep things moving along.
Incredibly, Datacamp.com provided to us free of charge for 6 months their entire catalog through their new Classroom Program used by places like Harvard and Berkeley, and it came with multiple Python for Data Science courses, and some fantastic advanced courses for when people graduated from the program! It was a life-saver!
To handle the different experience levels, we divided the two groups into white/yellow, and red/black, named after karate belts. The first group would obviously start on coding basics and then move on to Google’s Machine Learning Crash Course, while the senior devs would gloss over Python and move quickly into SVM’s and other tools in the Fast.ai course. Myself, the co-founder of Sugey, took the Intro’s, and Wes Roberts, the co-founder of Meter.ai, took the advanced team.
The advanced team, with Wes’s help, covered their materials quickly, and the participants meshed well with Wes’s advanced knowledge of the field. More then a few red/black belt participants commented on how they were surprised about how much they could cover with an in-person facilitator to answer questions. This highlights why it is more productive often to take training on site, with someone to answers questions quickly.
The white/yellow took more time to get into the Introduction to Python for Data Science course. Along the way, we answered a lot of questions about variables, methods, functions, lists, and more. Most everyone responded well to what must have been like drinking water from a firehose. By the end of the class, I was surprised not just by how fast new coders picked up the Python skills in the Datacamp course, but by the fact that the questions got better.
That’s a good sign.
The experience with the white/yellow belts brought up something I think about daily in my work, which is the effect of AI on the people around me.
I truly believe to open AI to the world, and break the group-think of the modern tech industry, we have to be willing to bring in people with no coding experience. We have to be patient, we have to be willing to give freely of our time, and we have to encourage.
The white and yellow belts did an incredible job doing the not-so-fun part of AI, and I believe many of these individuals can release their own simple AI and data science projects on their own by the end of the 10 weeks. From 0 to productive in that short amount of time is going to change lives.
As for the teams, we’re preparing for next week with training on how to download and run python code on people’s machines, reviews of last week’s materials, and an intro to matrix math. While time will tell how things turn out, we had eight additional people register after class for our next session after hearing good things from their friends.
I take that as another good sign.