Despite the incidents produced by Microsoft’s out-of-control experimental chatbot “Tay” which was caused by abusive manipulation from the internet community – the bot processed into its routines the information received by other Twitter users and formulated based on these its own, politically quite incorrect tweets1 – one should nevertheless consider the promising potential offered by this technology: an artificial intelligence which can learn by selected information and improve continously the complexity and quality of its output would be able to serve as a communicator and keeper of cultural knowledge.
The technology is not as new as it may seem: projects on learning AI programs are already in development for several years and delivering better and better results since the early 2000s. Beside numerous chatbot programs – some of them even made it to text almost indistinguishable from humans – the most prominent example is the “Watson” system from IBM which is already in experimental use as a kind of interactive encyclopedia in the medical sector. It is contrasting inserted data with its vast database and formulating responses for proposing medical treatment.2 The results are outstanding and proving that the development of AIs is not just an ephemeral hype.
Even if the currently programmed AIs do in fact not deserve the designation as intelligent due to their script-based operation modes, they have the potential to play starting with the next decade an important role not only in technical issues. They could also revolutionize the preservation of cultural heritage: if a conversational AI is loaded with the relevant and structured information on the history, culture and society of a population group – maybe even of a historical one which disappeared -, then it can serve as a communicator (in the group’s respective language and dialect or even in other languages) who by demand informs in an interactive way about the respective cultural heritage. This guardian of culture in form of a bot would make the access to information maybe even easier than an online encyclopedia and provide of course a much more vivid cultural experience. However: first of all there need to be made some major steps forward in AI development which is not, as must be stated, as immature as the now famous Tay bot, but nevertheless still needs a lot of development before being financially affordable for cultural projects. Some special attention by cultural scientists to this technology would surely be well invested.
- http://arstechnica.com/information-technology/2016/03/tay-the-neo-nazi-millennial-chatbot-gets-autopsied/ [↩]
- http://www.ibm.com/smarterplanet/us/en/ibmwatson/health/ [↩]