Chatbots are currently being boosted with the performance attribute AI.
However, most bots at present are being implemented in a relatively trivial
way. As a rule, certain keywords are scanned for on Twitter and Facebook,
on the basis of which predefined texts or text modules are then automatically played out. Somewhat more intelligent are systems that automatically detect relevant text findings on the Internet and then put them together
accordingly to form a post.
This automatic form of content curation is also discussed under the term
robot journalism. For the chatbots to be able to capture the posts accord-
ingly, the in the meantime significantly advanced processes of natural
language processing (NLP) which transform the running text into corre-
sponding semantics and signal words, are used.
Another approach is to connect the chatbots to knowledge databases. To
the user, chatbots seem to be “intelligent” due to their informative skills.
However, chatbots are only as intelligent as the underlying database.
Due to the advances in AI, chatbots can be by all means made more intel-
ligent in the future. AI-based chatbots learn largely independently from the
huge amounts of data available online and recognise question-and-answer
patterns that they use automatically in customer communication. The exam-
ple of Microsoft Tay mentioned shows, however, that the uncontrolled train-
ing of the bots by the community can lead to fatal consequences. The next
generation of AI-based bots must control and create the possible room for
communication.
With that, the degree of information supply is directly associated with the
degree of intelligence and automation of the bot. The present-day (usually
unintelligent) chatbots are fed the keywords, knowledge modules, texts and
rules of their developers/programmers. The more intelligent form of bots
obtains this information themselves from online sources and combines it to
form new content. The AI-based bots are also fed by the answers and reac-
tions of the users. The possibility of controlling thus also sinks for the infor-
mation used for learning.
Important food besides contents is also social signals such as likes and
followers. These enhance or reduce the impact of chatbots. This feedback
information can also come from other bots. So-called bot armies can make
contents and opinions go viral within a short time and thus automatically
set topic and agenda trends.
At the beginning, bots were able to answer simple, repetitive questions
that follow simple rules such as “What is the weather like today?” With the
advances in artificial intelligence and machine learning, bots can now take
over more demanding tasks. The idea of the bot goes back to the 1950s
when Alan Turing, a former researcher in computer intelligence, presented
a test to test the intelligence of machines. This is known to this day as the
Turing test and works as follows: If more than 30% of an experimental
group are convinced that they are having a conversation with a human and not with a computer, an intellectual power on a par with that of humans is
assumed of the machine.
Comments
Post a Comment