Our Dystopian Disconnect
WARNER BROS.

We may be getting a little ahead of ourselves sometimes and losing sight of why we embarked down this path of teaching machines to converse with us.


We laugh, mock and generally abuse the machines we’ve built and trained to converse when they say moronic things or repeat themselves. The critique is generally pretty negative without really reflecting why we’re on this journey.

We need this experimentation, pushing the boundaries and we are going to fail, many times over, then fail some more. This is how progress happens. But, if we step back and ask, what are the goal posts? What are we actually aiming for?

I believe many are stuck in this dystopian disconnect where we believe the end game is a scene from Blade Runner 2049, our evenings spent conversing with a machine trained to be our closest companion.

Tiernan Ray’s Why chatbots still leave us cold | ZDNet could use a little perspective from 60,000ft.

It’s unlikely that many of us will want to spend our days this way (I could well be wrong here).

Many of the tests carried out against the AI we’ve built emulate these open conversations assuming the machine has been encoded with the understanding of logic, context, knowledge and maybe emotion like ConvAI2. But, there’s still a long way to go before a general AI is going be any good. Let’s just accept this and stop the criticism.

My opinion, once they can understand sarcasm, we’ll be close to this pipe dream.

What if we shift the our expectations and the goal posts a little, narrow them so to speak? There’s no doubt in certain circumstances speaking your desires is far more frictionless than tapping, swiping and typing. Or even typing a single intent and having it interpreted, actioned and confirmed.

What we’re talking about is the next iteration of Siri, a smarter more capable assistant.

What if the goal posts were sending your friend $20 for dinner or changing your seat on the plane in a few steps. How about having your simple customer service request solved in minutes, rather than waiting on hold for 30 minutes?

If you take AI and narrow its focus it can become really quite good, not perfect but reasonable. It certainly won’t handle everything but for all the things it can how much time would it save you? How many of those frustrating interactions would the AI have made frictionless? Not to mention the people who weren’t helping you would have been helping those with the actual curly problems.

Maybe while we wait for our dystopian future, let’s reset the goal posts and focus on something where we can add some real value to our lives.

Maybe leave the romantic open ended conversations with machines for the year 2049?


Sebastian Pedavoli

Written by

Sebastian Pedavoli Co-founder and CEO @ Proxima.
Start the conversation

Try Iris free


Sign up free, no credit card required and become one of the many companies already using Iris to build conversations.

Try Free
Already use Iris? Log in