A few weeks ago, I was reading a few blog posts from two of the great minds at a16z about conversational interfaces being a fad and the prediction that no investor will look at AI startups in 2 years.
I thought I would add my three cents to the conversation.
See, from my vantage point, it's less about Alexa being the latest "it" fad put by amazon's geniuses to foster more ordering. It's more about another entry point and an additional way to interact with machines.
Let me explain. When I grew up the only method of interaction was the keyboard. Then came the mouse and the GUI. Then came the joystick. then (giant leap forward) the multi touch screen. Then another 10 years and we have consistently correct and realtime NLP, even when buried in harmonic noise. I went from LISP to Python, from 3 days to train a simple 3-layer perceptron to 3 mins. I've seen walled garden ecosystem plays evolve into API and platform businesses. From supply-side (Wal-Mart) to demand-side (Facebook) economies of scale. And everything in between.
All I see today is a proliferation of exciting ways to interact with machines that can only move us forward. Alexa and Echo are just the beginning. I've been absolutely mesmerized with Facebook's Oculus and the new Apple ARkit.
So my thoughts are It's not that in isolation conversational interfaces are narrow, in aggregate, added to all the other ways of we interact with machines, they add quite a lot. See, you can do something else with both your eyes and hands and still interact with echo and alexa. That's very powerful.
For example, say you're a doctor (I'm saying this because we have been working on it quite a bit lately) and you've got a very simple problem while you work - your hands and eyes are busy. Not your ears and mouth. So, from "Alexa please take notes about patient Chen" to "Alexa has my next patient arrived?" I see today simple and uselful ways to help small business owners, particularly in healthcare, get more efficient during their work day.
It's a tiny bit, but it always the sum of the little things.
Now about AI and the current hype: I did my first thesis on the recognition of spoken words using multilayer backpropagation perceptron in the early 90s. The math has been around since the 70s. So nothing new. What's innovative here is both the processing power becoming virtually free (which makes training somewhat easier) and the client-side approach (example iOS 11coreML) makes it super easy to put machine learning in any little gizmo with a DSP, harvest data, make much needed sense of this data very fast in real-time and send it to my smart cloud. That's pretty cool.
For what I'm doing right now, the combination of both those advances makes the idea of the "self-driving practice", which basically is a small healthcare business that is able to automatically 1. interact with clients (booking based on time inventory and client patterns), 2. manage supplies inventory (example medical supplies based on appointments and procedures), and 3. infers and optimizes time and fees for profit - totally doable today, something I could not do 3 years ago.
Today those two technologies, because they finally work well for real world applications, are enabling me to introduce, litle by little, leaps of efficiencies into healthcare - by automatizing the administrative costs.
So in conclusion, yes, from the standpoint of the VC looking for the first trillion dollar startup, what we see today is very much meh. It feels like innovation has stalled. I agree that given the excellence of iOS11 and ARKit/CoreML, we might see some Apple Glass showing up at the horizon, and that a well baked AR pair of shades will move us to another dimension of super cool stuffery. But the tiny little things add up and conversational interfaces, augmented with client-side AI help me solve real work-life problems for medical doctors today.
(PS. Here's a similar alternate view to a16z's posts discussed in the first paragraph, it's all good.)
My name's phil mora and I blog 2-3 times a month (it varies!) about the things I love: fitness, hacking work, tech and anything holistic. Since July 2015, I've been the Head of Product at Sikka. We're using big data, analytics, machine learning and AI to redefine medicine - our angle is the business of being a doctor. What we do is very, very cool. If you want to talk to me about that, you are more than welcome to contact me here or follow me on medium or twitter and chime in.
Join my newsletter:
I send my personal notes, usually for the week, with the links I found interesting and why they mattered. It's fun, I promise!