A World Without AI is Scary. Progress Isn’t. - Hypergiant

A World Without AI is Scary. Progress Isn’t.

02/18

TOPICS

#ARTIFICIAL INTELLIGENCE #DIVERSIFIED INVESTMENTS #PORTFOLIO MANAGEMENT #VENTURE CAPITAL #INDUSTRY

Written by John Frémont, Co-Founder and Chief Strategy Officer at Hypergiant



Introduction

There’s a war of opinions brewing in the conversation about AI. Between the linguists, the mathematicians, the Silicon Valley CEOs and the VCs that fund them, everyone wants to become a thought leader and share their opinion on the hot topic of Artificial Intelligence. While their motivations vary, they are fueled by the same thing: an unwarranted fear of AIs, their potential applications, and the possibilities of what it could do to humanity. And although these fears are reasonable to some degree, they are fundamentally flawed in context.

Consider the simple technical difficulties of having bluetooth earbuds, a smartphone, and a smart watch. Even if all three are made by Apple, the most profitable company in the world, getting the correct pair to interact without the interference of the third can be an exercise in frustration. Answering a call from your phone will somehow segue into your smartwatch, even though you intended to answer it from your fancy, pearly-white wireless headphones. Completing a proposal for a sales prospect from your phone will mistakenly not sync with the rest of your devices, even though you were told that it would. It seems that we can’t even get our smart devices, made by the same manufacturer, to talk to each other and deliver the user experience we were promised. So in essence, if we can’t optimize our current technological playing field, we shouldn’t really worry about some domineering Artificial Intelligence forcing humanity into subservience anytime soon. At this moment, it’s the least of our problems.

As with any new technology getting introduced into the market, there’s typically some confluence of fear, optimism, and indifference about what it’ll do for us. Artificial Intelligence is no different. But what many of us don’t realize is that AI is already operating in the world around us. It’s actively making our lives better by helping us expand our skill sets, keep our lights on, and discover treatments for the latest biological horrors, ultimately helping us become more productive, safer, and valuable over the course of our lifetime. And for some of the slow-moving, conservative industries we all know about, the potential longterm applications are enormous.

AI in Oil & Gas

While our dependence, as a species, on Oil and Gas has been decreasing, it will not go away entirely any time soon. So, wouldn’t it be better if we could be more efficient with how oil and gas exploration, production, and refinement is done? Consider the difficulties in oil and gas exploration, an estimated 60% of oil is beneath the sea floor, just finding it can be an exercise in danger for the workers. With autonomous robots, armed with computer vision packages, exploration of the seafloor as well as interpretation of the resulting data could greatly reduce the costs, both financial and in lives, of deep sea oil exploration. Now, having found the oil there is the difficulties in extracting it. Consider just the difficulty of drilling a hole six thousand feet into the earth you can’t see what the drill head is doing, you can’t see the material that makes up the sides of the well, even the slightest of a misangle could cause the drill head to increase the whorl and break. Here as well Artificial Intelligence can be used to help humans make decisions. Analysis of the soil composition via sensors in conjunction with positional sensors on the drill head allows an artificial intelligence to make decisions, or help a human make decisions by highlighting relevant data and learning a robust framework of predictive features that are, beforehand, not known to be related to drill problems. There is the actual refinement of the oil. There is a reason that petroleum engineers are paid as much as they are. Determining ways to improve the refinement of oil can be phrased as a reinforcement learning problem, which means the same techniques which allows artificial intelligences to beat the best chess players in the world can be used to make oil refinement more efficient. Finally, consider a world where intelligence in the machines can make vast corollaries far beyond human comprehension. Things like midstream pipeline management informing doctors on better mechanisms for fixing human circulatory issues. It’s certainly not out of the realm of possibility.

AI in Weather

The argument about climate change is even older than the argument of AI ethics. Why has it not be settled already? One, fundamental, reason is that the models are really complicated. Think about this, the sun heats the atmosphere, which heats the water and the land, which causes air and water circulations which moves hot and cold air and water. However, these interactions are modified by stationary affects like mountains, but also temporary affects like one current of hot air running into the side of a current of cold air. This means that effects are all but impossible to predict without massive computation, if you know the state of every air and water flow on the globe at the same time, their temperature, speed and direction. But, we do not know all of these, in fact, we barely know any of them. Rather than being upset when a weather prediction is wrong we should be in awe that they are correct as often as they are! It was not that long ago in history when sailors and farmers used very anecdotal references to attempt to predict weather patterns — “Red sky at night, sailors delight, red sky in the morning sailors take warning”. In a blink of the eye, in terms of history, we have the gall to make global predictions of weather patterns! Think how much better those predictions could be if we could incorporate more information, more patterns, more correlations into our models! However, once we get to computations with that range of complexity it no longer lies within the realm of traditional computation. At that stage we are now seeking high level correlations between diverse features: How did the supermoon over the North Sea and the El Nina in the South Pacific combine to affect weather in Panama? These types of questions can only be answered with the abstraction that is part of Artificial Intelligence. In particular, in 2015 Artificial Intelligence Researchers at Microsoft using historical weather data made predictions that were more accurate than any of the traditional methods by a noticeable amount. Since then Artificial Intelligence has made much headway into the meteorological community, to the point that if you watch the weather prediction on your local news station, it was probably computed, in part, by an Artificial Intelligence.

AI in Automotive

According to the CDC more than 33,700 people died in 2013 in automotive accidents. In fact by comparing like to like it is safer to fly than to drive. In the past few years we have seen the emergence of automatic braking of cars, and more sophisticated real time image classification of cars in video. These are both signs that there is interest in developing, and applications for, Autonomous Vehicles. Cars driving themselves, with their ability to accurately judge the situation independent of weather, lighting, or impaired human judgement has been the dream state of science fiction for years: Set the autopilot, and let the computer fly the spaceship! The humanitarian aspect alone, the mobility these would present to the physically or mentally handicapped, the reduction in loss of life, these make them worthwhile. Or consider the difficulty in long distance driving. Who, when driving long distance, hasn’t had at least one moment of shock when they realized they were not paying attention, got distracted, or started to nod off? Autonomous diving, in their simplest form, could be used to augment a human’s driving skills, so if they behave in a manner similar to that the vehicle gets their attention or safely drives the vehicle to the breakdown lane. It’s ominous to those of us who grew up in an era where cars meant freedom, but does it not simply open up another type of freedom? Even more ominous to the car manufacturers who are desperately trying to help carve a path in this revolution they’ll ultimately move beyond units off the line to more advanced implementations to the infrastructure. There will be winners and there will be losers in the car maker race but ultimately, humans win.

AI with Personalization

We have discussed how Artificial Intelligences improve, or offer improvements, in the industries around us, ranging from making oil and gas more affordable to literally life saving accident reduction. What does all of this look like from an individual’s perspective? Artificial Intelligence gives us a chance to regain our humanity. In the first industrial revolution individuals were torn from their ancestral homes, their families, and the lives they had known by an emerging economy that none of them understood. That sent shockwaves throughout human society dehumanizing the workers in the eyes of the factory owners, this dehumanization was only strengthened in the development of heavy industry in the second industrial revolution. This cultural damage has left us reeling, but, AI could be used to help fix that. Artificial Intelligence could be used to help buffer us, and our families, from the vagaries of corporations like Google, Amazon, and Apple. Artificial Intelligences offer ways to streamline life in ways that allow us to spend more time with our families, our children. While Artificial Intelligences can be abused, as is often the case with social media and Artificial Intelligences in that area, the potential is there to make family life no longer something which happens as an afterthought of school and work but rather, once again, as a centerpiece of our lives.

Conclusion

Finally, and a bit of digression, Hypergiant doesn’t prefer “Artificial Intelligence” as a useful term — it’s been so bastardized by the mainstream conversation that it lacks specificity and sets people’s mind to things of science fiction. We ascribe to the nomenclature of “Machine Intelligence” as there is nothing artificial, after all, about designing technology to make people’s lives better, or to make technology more intelligent. Thus, rather than the boogeyman of artificial intelligence from science fiction, we design and build Machine Intelligences to help humans in their daily lives.

This term better defines the Fourth Industrial Revolution — one which Hypergiant is helping usher in. A world without AI is scary, but progress isn’t.