In the first article of this AI five-parter, we broke Artificial Intelligence into three component parts: perception, cognition and action, acknowledging that the term AI is often, and wrongly, shorthand for self-determining robots and near-sentient automatons.
The glossy AIs you see in movies and mags – they might be out of reach for today, but there's still huge opportunities to gain a competitive edge with AI and its three component parts.
First up, let's open the doors of perception.
Perception in AI is the collecting and processing of data from the surrounding environment. Examples include image and object detection, as used in machine vision; speech recognition, the kind deployed in call centres; and location sensing, the likes of GPS, LIDAR and the Radar systems used in self-driving cars.
Processing natural language is a challenge in AI development. There's inherent problems with language: it's irregular, it's context dependent, and Stephen Fry is just about the only CPU who can use it perfectly.
More than just trading, entity extraction and sentiment analysis can predict any outcomes that are driven by public perception.
However, if machines are ever able to master seamless human conversation, the possibilities are immense. That's why there's a smorgasbord of tools in the field of Natural Language Processing (NLP) as we attempt to build that future.
In NLP, there's entity extraction algorithms to identify names and nouns. There's sentiment analysis to gauge texts' positivity and negativity. Topic analysis can establish themes and word sense disambiguation helps to figure out words' meanings in context.
Singularly and joined up, these techniques are in use today. Entity extraction and sentiment analysis work hand in hand to gather news and to gauge opinion on events. Not only are they great allies in day trading, they're used in the fight against negative public perception. United Airlines take note.
More than just trading, entity extraction and sentiment analysis can predict any outcomes that are driven by public perception. At Aelm, we conducted a cursory sentiment analysis on Twitter and successfully predicted the outcomes of Brexit and the US election well ahead of time.
Some view NLP tools as the blueprint for customer-facing AI.
In deploying topic analysis, organisations can ensure their social media messages align with corporate values, plus it can help a business to identify opportunities; by exposing language and opinion trends.
Now that financial service organisations are required to hit oppressive compliance targets, NLP can help them ensure assets, like customer documentation, are readable, clear and understandable. When it comes to FS marketing, NLP can be a real blessing in an era where compliance is more important than creativity.
Oh and we can't chat NLP without chatting chatbots: conversational agents that understand us. Sexy examples include Alexa, Siri and Cortana, unsexy examples include your gas suppliers' online customer service.
Because chatbots can, within limits, autonomously handle basic tasks and enquiries, they make enormous fiscal sense, not least because they free staff from the mundane to apply skills elsewhere.
Some view NLP tools as the blueprint for customer-facing AI. But because AIs learn their language from humans – whether that's the coders programming them, or the sentiment of mass audiences – their learning algorithms can compound biases: AIs can learn human unpleasantries as was reported last week.
But imperfections and unhelpful headlines aside, NLP offers tools that can drive real business value right now.
Find out more about Aelm at www.aelm.com.
