11

Machine Learning with Barbie

Grace Miller

As a little girl dolls have always been a large part of my childhood. I would spend most of my hours in my room playing with them and living in their imaginary would. Like many kids I would give them all names and act out their lives for them as if they were real. At times I would even incorporate myself into their lives and talk about the things I enjoy. However, this relationship was very much on sided. As technology is advancing the toys that parents are giving to their children are becoming more and more interactive. One of the most recent advances in toy technology is Hello Barbie. The newest version of Barbie features a small microphone hidden in her necklace that records what’s being said to her. There is a “listen” button that the child can press which starts the recording process of the conversation. This information is then transmitted to the servers at ToyTalk where the information is analyzed, and appropriate responses are sent back to barbie to deliver. In order to make it seem realistic all of this occurs within a second allowing for a smooth interaction between Barbie and the children that are playing with her. Barbie is also then able to remember the information given to her and recall at a later time in a conversation days or weeks later. This kind of machine learning is very similar to Cynthia Braezeal’s Leonardo. He is an interpersonal robot that is able to learn through interactions with humans and produce appropriate reactions. Both Barbie and Leonardo are able to pick up on the semantics in an individual’s voice in order to understand what is being said to them and respond accordingly. Barbie has all potential conversations mapped out and is able to choose one of the 8,000 lines of dialogue she has available. The voice of Barbie is also pre-recorded by an actor allowing her to feel more human-like and relatable. She is also updated in order to keep up with the latest trends and pop culture.

Although Hello Barbie is quite impressive in its speed and ability to respond to a vast number of topics, she does have quite a bit of limitations that keep her from being an Intelligent machine. Even though Barbie is able to learn new things about the children that are playing with her and recall in order to have smooth interactions she is not truly gaining new intelligence. As discussed in Andy Clark’s Symbol Systems chapter the physical symbol system hypothesis claims that intellectual performance can be explained by symbol manipulation in both humans and machines (PSS). This means patterns are taken in, combined into structures and manipulated to produce new expressions. Barbie is able to fulfill the three commitments of the PSS to a certain degree as she is able to use symbolic coding in order to store information, search a symbolic problem space and seemingly produce deliberative thought. I would argue that Barbie does not necessarily have deliberative thought as she cannot start conversations on her own although she does learn as she is being spoken to. In many ways she is similar to the chatbots we have discussed in class. She is able to respond when spoken to and when she is faced with a topic she doesn’t know how to respond to she has default answers that give away that she is not fully installing. This might not be an issue for children because they might not read into Barbie’s responses with as much criticism as someone who is looking to see how well she really performs as AI. This can all be seen in the ways she is able to store the information given to her and recall at a later date. Hello Barbie however, does not display true intelligence in the sense that it does not have its own thought process. She is however the kind of interactive robot that Braezeal envisioned for humans to connect with. We are able to interact with her as we wish but we are not overwhelmed by her presence and it is not creepy to us since we are used to interacting with our smartphones in a similar manner.

Hello Barbie is equipped with speech recognition software that uses natural language processing to sort through the information given to her. This software is definitely the backbone of Hello Barbie and allows her to interact smoothly. In the mindware as software reading it is explained how the brain is very much  However Barbie does not have the same brain capacity as humans in the sense that it is not as malleable.