About
This project is about injecting emotions into AI (and a chatbot to start with). We are not talking about detecting the human emotions of the user (sentiment analysis), but about making the bot and the AI algorithms emotive.
Our system is based on a hybridization of ML and OR and our approach is unique. Contrary to the traditional ML approach, we don’t need to gather lots of data and our system can change its mood on its own.
Why using only ML for emotions is not a good idea
The “classical” way to inject emotions into a conversational system is to mimic or simulate the emotions that the system can extract from a lot of data involving emotions. This approach is doomed to fail, for the following reasons:
- You need to gather a lot of data. If the system encounters a new context for which there is no data available, the system will react poorly.
- Even if you have the corresponding data available, it will be biased — especially when we deal with emotions. Biases are unavoidable in data.
- The system cannot autonomously change its emotions based on the interactions, except in a few basic scenarios. Points 1 and 2 apply here too. The only way such systems could react satisfactorily would be to have all possible data in the universe — and even then, they wouldn’t be able to react to new situations.
- Such systems cannot be controlled, i.e. they could go berserk. With ML alone there is no way to control the final answers. This is not just theory: in practice chatbots — emotive or not — quite often spit out nonsensical answers, not to mention that they can be rude and lack the right emotion.
Our prototype, Philia, takes the other road.