A View from the Future — CB Insights Innovation Summit Day 1
Yesterday, Kevin Stevens and I had the great pleasure of attending the first day of the CB Insights Innovation Summit in Santa Barbara. Going back over my notes, it will take me weeks to fully digest all the ways in which I’m updating my worldview, and this will likely yield a few long-form pieces over the coming weeks. But for today, you get a few very brief insights from a day full of trying to conceive what the future will be like, then reverse engineering the process of getting there.
A few big takeaways:
UI 2.0
Voice is the clear leader over gesture when it comes to next generation UI. Several disadvantages will hamstring gesture control (sorry BMW), but the principal issue that seems clearest to me is that of fidelity. I simply have more finite ways of communicating my desires to Alexa. One thing that I believe remains as-yet unsolved is the issue of discovery over these new UI paradigms.
Jeremy Liew of Lightspeed Venture Partners pointed out some valid issues about how voice interaction may result to a pull model of interaction with our devices. It’s much easier for me to browse the app store to discover what my phone can offer than to ask Alexa to offer me a giant list of all the skills it can provide. As it stands, new features for Alexa and Google Home are largely still discovered on a mobile screen. I’m not suggesting that this is an unsolvable problem, but so far it offers a strategic foothold for Google and Apple in this market as they already own significant “pocketshare.”
AI 1.0
We may be some way from any imminent AI threat of the kind that keeps Elon Musk awake at night, but we are now moving from a model of “machine-assisted humans” to “human-assisted machines.” Two forces are at play here. First, you previously would use an Excel-spreadsheet to investigate a single query. Now, humans are needed to tune the parameters of machine learning, but feeding the machine data can now offer a plethora of (hopefully) useful outputs.
But more immediately and widely significant is the rise of artificial specific intelligence, particularly as it relates to bots. Chat bots are now capable of reasonably natural language interaction, allowing them to hold a basic conversation with a customer, ask questions, etc. A bot can then forward the user’s answers to a specialist in that field. While, due to the implicit risks, this scenario is at least a few years away, you can easily imagine a chat bot asking a sick patient about their symptoms. It then takes the patient’s answers and forwards those to the relevant tele-doc specialist. A more immediate application is customer service, where a chat bot can begin by answering fundamental questions, but may push the interaction to a human customer service agent down the line if more interactive troubleshooting is required. These capabilities are going to remake low-to-medium skill industries beginning now, moving up the risk and complexity chain over time.
Cybersecurity in IIoT
I won’t go deep on this one right now, but Michael Dolbec from GE Ventures made some comments about the precautions they’re taking relative to cybersecurity for IoT devices. Despite the awareness most of the big players show of this issue, I still think it’s likely under appreciated. In the next 20 years we will deploy more sensors into the world than ever before, and these new sensor networks will influence outcomes in the physical world in a way that is likely difficult to presently appreciate. The implication here is that, viewed from a deterministic sense, a cybersecurity threat to IoT devices could offer much more dire consequences than something analogous to the Yahoo hack.
Going back to the seminal pieces of research around industrial accidents, Charles Perrow’s research around so-called “normal accidents” in essence states that as the complexity of a system increases, the potential interactions within this system increase in numerous and nearly impossible to anticipate ways. This yields an outcome where system failures are not just more likely, but expected. If ever there was a use case that matches the academic research, it would be deploying hundreds of millions to billions of sensors from many providers across an innumerable cross section of geographies and industries, and empowering most of those sensors to talk to each other and influence the physical world.
This brings me to Chamath Palihapitiya’s insights around Silicon Valley’s traditional views toward regulation, suggesting that their disdain is largely driven by naïveté. Regulation serves more as a belt than like handcuffs. Regulation isn’t always perfect, but its intent is to maintain consumer trust and limit negative outcomes, not constrain innovation. Entrepreneurs and large companies should expend more effort trying to work with regulators, not around them. This is certainly true as we build our rules around industrial safety in the age of connected devices.
I believe in the promise of IoT, and that it will radically improve our world over the next few decades. But there almost certainly will be teething problems, and we should anticipate that this outcome is nearly inevitable. If we don’t prepare ourselves for some amount of friction, I fear humans will be too quick to pull back our trust from machines that could ultimately have a significant positive effect on humanity.
Written by Jonathan Crowder. Edited by Christian Poutge.
Learn more about DJI Innovation’s latest MAVIC PRO, DJI’s portable yet powerful personal drone by watching the video below:
SUGGESTED READING:
