ELIZA and Chatbots

sheldonI first encountered a chatterbot, it was ELIZA on the Tandy/Radio Shack computers that were in the first computer lab in the junior high school where I taught in the 1970s.

ELIZA is an early natural language processing program that came into being in the mid-1960s at the MIT Artificial Intelligence Laboratory. The original was by Joseph Weizenbaum, but there are many variations on it.

This was very early artificial intelligence. ELIZA is still out there, but I have seen a little spike in interest because she was featured in an episode of the TV show Young Sheldon. The episode, "A Computer, a Plastic Pony, and a Case of Beer," may still be available at www.cbs.com. Sheldon and his family become quite enamored by ELIZA, though the precocious Sheldon quickly realizes it is a very limited program.

ELIZA was created to demonstrate how superficial human to computer communications was at that time, but that didn't mean that when it was put on personal computers, humans didn't find it engaging. Sure, kids had fun trying to trick it or cursing at it, but after awhile you gave up when it started repeating responses.

The program in all the various forms I have seen it still uses pattern matching and substitution methodology. She (as people often personified ELIZA), gives canned responses based on a keyword you input. If you say "Hello," she has a ready response. If you say "friend," she has several ways to respond depending on what other words you used. Early users felt they were talking to "someone" who understood their input.

ELIZA was one of the first chatterbots (later clipped to chatbot) and a sample for the Turing Test. That test of a machine's ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human, is not one ELIZA can pass by today's standards. ELIZA fails very quickly if you ask her a few complex questions.

The program is limited by the scripts that are in the code. The more responses you gave her, the more variety there will be in her answers and responses. ELIZA was originally written in MAD-Slip, but modern ones are often in JavaScript or other languages. Many variations on the original scripts were made as amateur coders played around with the fairly simple code.

One variation was called DOCTOR and was made to be a crude Rogerian psychotherapist who likes to "reflect" on your questions by turning the questions back at the patient.  This was the version that my students when I taught middle school found fascinating and my little programming club decided to hack the code and make their own versions.

Are chatbots useful to educators?  They have their uses, though I don't find most of those applications to be things that will change education in ways I want to see it change. I would like to see them used for things like e-learning support and language learning

If you want to look back at an early effort, you can try a somewhat updated version of ELIZA that I used in class at my NJIT website. See what ELIZA's advice for you turns out to be.

 

Is Our Group a Learning Community, Learning Circle or Community of Practice?

Though there are differences, you will often find the terms Learning Community, Learning Circle and Community of Practice used interchangeably. They are all groups of individuals who learn from each other, and with each other, on an ongoing basis with the goal of improving their work. 

Like any network of people, communities of practice are generally self-organized by people who share common work practice. As with the other labels, any of these relationship groupings have a desire to share what they know, support one another, and create new knowledge for their field of practice.

But communities of practice (CoP) differ from networks in that they are intended to be "communities" in which people make a commitment to be there for each other. They should participate not just for their own needs, but to serve the needs of others.

A CoP is very "open source" with a commitment to advance the field of practice and to make their resources and knowledge available to anyone, especially those doing related work.

A learning circle is a highly interactive, participatory structure for organizing group work. The goal is to build, share, and express knowledge though a process of open dialogue and deep reflection around issues or problems with a focus on a shared outcome.

Online learning circles take advantage of social networking tools to manage collaborative work over distances following a timeline from the open to close of the circle. Learning circles usually have a final project or goal which collects the shared knowledge generated during the interactions. Learning circles are a way to organize learning in global projects. They are also being used in Massive Open Online Courses (MOOCs).

But again, there is crossover with these terms. I have even seen articles about "Creating a Community of Practice Using Learning Circles

Almost anyone can facilitate a learning circle, whether it is a single learning circle in your home or multiple circles across a an organization like a university or library system. 

Teaching With 40 Year Old Software

I read an article that mentioned that someone teaching game design was using the old game "The Oregon Trail" as a simple example of game design. I felt a little wave of nostalgia for that computer game that I used with middle school students in the late 1970s on Apple IIe computers.

What can we teach with 40-year-old software?

The game was developed in 1971 and produced by the Minnesota Educational Computing Consortium (MECC) in 1974. My school subscribed to MECC and received many software packages on the big 5.25 very floppy disks which we could duplicate.

The original game was designed to teach about the realities of 19th century pioneer life on the Oregon Trail. The single player is a wagon leader guiding a party of settlers from Independence, Missouri, to Oregon's Willamette Valley on the Oregon Trail via a covered wagon in 1848. 

But many teachers used it in other ways. In those early day, just teaching students to use the computer and navigate a game was a learning experience. I knew teacher who, like myself, used it was a way to teach cooperation by having players work in pairs or teams and justifiable arguing about choices was encouraged.

I used the game as an example when teaching literature as away to discuss the consequences of actions (draw branching diagram here).
 

Looking at the game again today via one of the several emulators available online (such as https://archive.org/details/msdos_Oregon_Trail_The_1990 and https://classicreload.com/oregon-trail.html), it seems about as primitive in its graphics as it did back in 1975 in my classroom. But it worked. My homeroom students enjoyed playing it just for gaming fun, and I was able to incorporate the decision-making aspects into lessons. I taught English, not social studies, and was less interested in the historical aspects of the game. I did use it briefly in an interdisciplinary manner with a social studies teacher, but having students do research into the real Oregon Trail and that period seemed to kill interest in the game itself. 

Apple IIe screenshotIt was one of the most successful games of all time and “The Oregon Trail” was inducted into World Video Game Hall of Fame in 2016. If you played it a few times, many of its screens are probably etched into your memory. I recall entering my real family members' names into the game the first time I played, and then sadly dysentery them "die" along the trail - probably from dysentery. It had game play moments (like hunting buffalo) and simple animation, but it was mostly text and so involved a lot of reading.

I would have my students work in small groups and map the game both on a real map of the trail, and then later on a decision tree style "map" of the game's options.

For me, the strength of the game in the classroom was in understanding how decisions could change the game's outcomes and their traveler's fates.

I recall that students would argue about the design. They didn't like the random things that would happen, such as a fire in the wagon destroying objects that were worth game points. But that also worked its way into my discussions with them of literature. Things happen in novels - and our lives - that seem random and out of our control, and they have consequences.

The other software that I used back then which was more sophisticated (though not graphically) was made by Tom Snyder Productions. I met Snyder at an educational conference and we talked about his Decisions, Decisions series. The series focused on the best aspects of what I was using in "Oregon Trail." The series included products on politics and the environment and came with printed material to supplement the games, so "research" was easy and necessary to play well.

I had no luck finding online what happened to Snyder and his company. It seems to have been consumed by Scholastic, though the link I found was a dead end.  I did find something on Amazon, but it doesn't seem that the series was continued or updated recently. It could easily be an online or mobile game. 

Can we use old software to teach new skills? Absolutely. Though these software packages seem crude by today's standards, they are also "classic" curiosities. I haven't taught secondary school students since 2000, so my sense of what is acceptable is lost. Certainly some of these games, or similar decision-tree kinds of games are a very viable classroom tool at all grade levels K-20. Maybe someone has already updated them or created new versions. If not, there is an opportunity.

     

Bleeding Edgy Deep Learning

Deep learning is a hot topic right now, but it is not lightweight or something I would imagine learners who are not in the computer science world to take very seriously. But I stumbled upon this video introduction that certainly goes for an edgier presentation of this serious subject and obviously is trying to appeal to a non-traditional audience.

That audience would be part of what I refer to as both Education 2.0 and also that segment of learners who are The Disconnected.  I see these disconnected learners as a wider age group than "Millennials." They are the potential students in our undergraduate and graduate programs, but also older people already in the workplace looking to move or advance their careers. The younger ones have never been connected to traditional forms of media consumption and services and have no plan to ever be connected to them. And that is also how they feel about education. You learn where and when you can learn with little concern for credits and degrees.

The video I found (below) is an "Intro to Deep Learning" billed as being "for anyone who wants to become a deep learning engineer." It is supposed to take you from "the very basics of deep learning to the bleeding edge over the course of 4 months." That is quite a trip. 

The sample video is on how to predict an animal’s body weight given it’s brain weight using linear regression via 10 lines of Python.

Though the YouTube content (created by and starring Siraj Raval) is totally free, he also has a partnership with Udacity in order to offer a new Deep Learning Nanodegree Foundation program. Udacity will also be providing guaranteed admission to their Artificial Intelligence and Self-Driving Car Nanodegree programs to all graduates. 


Is this a good marketing effort bu Udacity? Will it reach new and disconnected learners? Will they simply use the videos and resources to learn or make that connection to some kind of degree/certification that might tell an employer that they know something about deep learning? I don't have the deep learning program that can predict that. I'm not sure it exists. Yet.

RESOURCES

This is the code via GitHub for "How to Make a Prediction - Intro to Deep Learning #1' by Siraj Raval on YouTube

This lesson uses simple linear regression. "Simple" is a relative term here, as many people would not find it simple, as in "easy." It is a statistical method that allows us to summarize and study relationships between two continuous (quantitative) variables. This lesson via Penn State introduces the concept and basic procedures of simple linear regression.

You might also want to look at this tutorial on the topic via machinelearningmastery.com.

Humans Learning About Machines Learning

GoDeep learning might sound like that time when we get really serious about what we are thinking about, and go deeper into the subject and learning. But it is not about the human brain. It is about machine learning. Also known as deep structured learning or hierarchical learning, it is part of the study of machine learning methods. It is about machines getting smarter on their own as they complete tasks.

The theories do look at biological nervous systems as models. Neural coding attempts to define a relationship between various stimuli and associated neuronal responses in the brain The terms used are many. Deep learning architecture, deep neural networks, deep belief networks and recurrent neural networks are all labels used in computer vision, speech recognition, natural language processing, audio recognition, social network filtering, machine translation, bioinformatics and drug design. That means a machine is producing results instead of human experts.

Google's artificial intelligence software, DeepMind, has gotten a fair amount of press coverage. It has the ability to teach itself many things, including how to walk, jump, and run. In the press, it will defeat the world's best player of the Chinese strategy game, Go, but deep learning is more serious than that.

You can take a free, 3-month course on Deep Learning offered through Udacity, taught by Vincent Vanhoucke, the technical lead in Google's Brain team.

Machine learning is a fast-growing and exciting field of study and deep learning is at its "bleeding edge. This course is considered to be an "intermediate to advanced level course offered as part of the Machine Learning Engineer Nanodegree program. It assumes you have taken a first course in machine learning, and that you are at least familiar with supervised learning methods."