The twitter-bot tutorial was great. I had no knowledge of what a twitter-bot was until that lesson, as I had never worked with twitter and had never looked into it very far either. The range of possible uses of the twitter bot are fascinating to me, as you can essentially make them do whatever you would want, as long as you have the coding skills to pull it off. That was, for me, one of the minor negatives of the workshop. Despite having gone through the majority of the code academy tutorial and having learned a decent amount of coding, i still found that I did not know near enough to make a bot that would function anywhere near a way that I would want it to.
The possibilities that I found most intriguing about the bot itself were not so much in posting to twitter, but instead its ability to do a significant amount of research for a user with a very small amount of effort necessary. The search function that can be programmed into a bit, which I attempted to toy with a bit, was extremely intriguing. It could be a very useful and easy way to immediately search for and collect data on certain topics as they appear in tweets, such as frequency, positive or negative connotation, etc. I am sure that there are plenty of people out there currently exploring these functions in interesting ways, but on a personal level it would be worthwhile to be able to manipulate and use bots in this way.
Modulus,
I agree that the search function is much more interesting than simply posting an endless barrage of text, although I myself wasn’t even close to running searches by the end of the workshop. The search function is the key to creating a bot that can actually dialogue with users–currently, most chatbots use keyword searches to interpret what the user says and produce a related response. That’s why they often seem to respond to some words in your input but not others. However, I’m skeptical about whether a bot could ever pass a Turing test in this way. As Matt noted in the workshop, there are already customer service bots that talk to customers, but it’s usually easy to tell from the timing and the content of their responses that you’re not talking to a human. Since tweaking tense or word order can completely change the meaning of a sentence, keyword searches are good enough for responding to hashtags, but I don’t think they’ll ever be able to simulate real conversation.