I read about Woebot and decided to try it out for myself. Without a specific use case, I just decided to just engage with the chatbot to see what that felt like. The whole interaction is tightly scripted, so your responses back to the bot can only be one of the available options that it has been coded for. The moment I entered free form text, its ability to converse "intelligently" plummeted. It tried to get me back on track by prompting me to choose a word that best described what I was feeling. It took less than two minutes to throw it off its game and turn the interaction from mildly interesting to just pointless. The Wired story about Woebot had this to say about it:
In some ways, a CBT chatbot is the ultimate manifestation of that philosophy. “Woebot is a robot you can tell anything to,” says Darcy. “It’s not an AI that’s going to tell you stuff you don’t know about yourself by detecting some magic you’re not even aware of.” Woebot only knows as much as you reveal to it—and it can only help as much as you decide to help yourself.
It is not really possible to reveal much to Woebot it seems. Your freeform messages won't make immediate sense to it but over time and with wider use adoption it is likely to make better sense of it than it does now. The more data to train the bot with the better and you could look at yourself helping out a good cause in this way - supplying it training data. As far as immediate benefits to yourself - not much to write home about. Based on my limited interaction, it seemed like a chat with Woebot was just as illuminating as an iChing reading online. If you test it with a large number of questions, at some point the response seems plausible even profound. As with other slick AI snake-oil offerings Woebot has garnered a great deal of press. The name itself I found clever but rather unfortunate.
In some ways, a CBT chatbot is the ultimate manifestation of that philosophy. “Woebot is a robot you can tell anything to,” says Darcy. “It’s not an AI that’s going to tell you stuff you don’t know about yourself by detecting some magic you’re not even aware of.” Woebot only knows as much as you reveal to it—and it can only help as much as you decide to help yourself.
It is not really possible to reveal much to Woebot it seems. Your freeform messages won't make immediate sense to it but over time and with wider use adoption it is likely to make better sense of it than it does now. The more data to train the bot with the better and you could look at yourself helping out a good cause in this way - supplying it training data. As far as immediate benefits to yourself - not much to write home about. Based on my limited interaction, it seemed like a chat with Woebot was just as illuminating as an iChing reading online. If you test it with a large number of questions, at some point the response seems plausible even profound. As with other slick AI snake-oil offerings Woebot has garnered a great deal of press. The name itself I found clever but rather unfortunate.
Comments