Amazon Alexa has moved on in leaps and bounds since it was first launched in 2014, but it has largely been on the basis of simply asking it something and getting a response. Following on from the addition of conversational speech during the summer, the latest upgrade to the service is set to change all that with the addition of “latent goals”, which lets Alexa infer what you actually want before you’ve asked.

One example from the Amazon blog could be making a cup of tea. You might start with a question like “How long should I steep Earl Grey tea for?”. Alexa will tell you but could follow it up with questions about your ‘latent’ goal - the one you haven’t mentioned. It could say “Shall I boil the kettle?” if you have one integrated. It might suggest it sets a timer for the correct steepage time.

Although this sounds like a logical progression, it represents a huge leap in Alexa’s AI abilities. One of the challenges of artificial intelligence is that, unlike humans, neural networks don’t have a context or learned experience to draw from, and training it to draw logical conclusions like “Ah - he must be making a cup of tea” has been a mammoth effort by Amazon’s engineers - not least of all because not all requests have a latent goal, and the last thing you want is Alexa trying to draw conclusions that aren’t there.

US English users of Alexa devices can already use the new functionality, but it will really come into its own as individual Alexa Skills are updated to work better with it. This can be done with the Name-Free Interaction Toolkit, which allows you to identify ‘hooks’ for the latent-goal engine. As Alexa learns from “on-the-job” training based on the millions of interactions she has, and as more adapted apps come online, you should find she’s more helpful than ever.