Skip to main content

Home/ GAVNet Collaborative Curation/ Group items tagged machines

Rss Feed Group items tagged

Filter: All | Bookmarks | Topics Simple Middle
Bill Fulkerson

Anatomy of an AI System - 1 views

anatomyof.ai

AI amazon

shared by Bill Fulkerson on 14 Sep 18 - No Cached
  • Bill Fulkerson on 14 Sep 18
     
    "With each interaction, Alexa is training to hear better, to interpret more precisely, to trigger actions that map to the user's commands more accurately, and to build a more complete model of their preferences, habits and desires. What is required to make this possible? Put simply: each small moment of convenience - be it answering a question, turning on a light, or playing a song - requires a vast planetary network, fueled by the extraction of non-renewable materials, labor, and data. The scale of resources required is many magnitudes greater than the energy and labor it would take a human to operate a household appliance or flick a switch. A full accounting for these costs is almost impossible, but it is increasingly important that we grasp the scale and scope if we are to understand and govern the technical infrastructures that thread through our lives. III The Salar, the world's largest flat surface, is located in southwest Bolivia at an altitude of 3,656 meters above sea level. It is a high plateau, covered by a few meters of salt crust which are exceptionally rich in lithium, containing 50% to 70% of the world's lithium reserves. 4 The Salar, alongside the neighboring Atacama regions in Chile and Argentina, are major sites for lithium extraction. This soft, silvery metal is currently used to power mobile connected devices, as a crucial material used for the production of lithium-Ion batteries. It is known as 'grey gold.' Smartphone batteries, for example, usually have less than eight grams of this material. 5 Each Tesla car needs approximately seven kilograms of lithium for its battery pack. 6 All these batteries have a limited lifespan, and once consumed they are thrown away as waste. Amazon reminds users that they cannot open up and repair their Echo, because this will void the warranty. The Amazon Echo is wall-powered, and also has a mobile battery base. This also has a limited lifespan and then must be thrown away as waste. According to the Ay
Steve Bosserman

Which Industries Are Investing in Artificial Intelligence? - 0 views

priceonomics.com/es-are-investing-in-artificial

artificial intelligence machine learning investment future trends

shared by Steve Bosserman on 29 Nov 18 - No Cached
  • The term artificial intelligence typically refers to automation of tasks by software that previously required human levels of intelligence to perform. While machine learning is sometimes used interchangeably with AI, machine learning is just one sub-category of artificial intelligence whereby a device learns from its access to a stream of data.When we talk about AI spending, we’re typically talking about investment that companies are making in building AI capabilities. While this may change in the future, McKinsey estimates that the vast majority of spending is done internally or as an investment, and very little of it is done purchasing artificial intelligence applications from other businesses.
  • 62% of AI spending in 2016 was for machine learning, twice as much as the second largest category computer vision. It’s worth noting that these categories are all types of “narrow” (or “weak”) forms of AI that use data to learn about and accomplish a specific narrowly defined task. Excluded from this report is “general” (or “strong”) artificial intelligence which is more akin to trying to create a thinking human brain.
  • The McKinsey survey mostly fits well as evidence supporting Cross’s framework that large profitable industries are the most fertile grounds of AI adoption. Not surprisingly, Technology is the industry with highest AI adoption and financial services also makes the top three as Cross would predict.Notably, automotive and assembly is the industry with the second highest rate of AI adoption in the McKinsey survey. This may be somewhat surprising as automotive isn’t necessarily an industry with the reputation for high margins. However, the use cases of AI for developing self-driving cars and cost savings using machine learning to improve manufacturing and procurement efficiencies are two potential drivers of this industry’s adoption.
  • ...2 more annotations...
  • AI jobs are much more likely to be unfilled after 60 days compared to the typical job on Indeed, which is only unfilled a quarter of the time. As the demand for AI talent continues to grow faster than the supply, there is no indication this hiring cycle will become quicker anytime soon.
  • One thing we know for certain is that it is very expensive to attract AI talent, given that starting salaries for entry-level talent exceed $300,000. A good bet is that the companies that invest in AI are the ones with healthy enough profit margins that they can afford it.
Steve Bosserman

Opinion | It's Westworld. What's Wrong With Cruelty to Robots? - 1 views

www.nytimes.com/...conscious-robots-morality.html

ethics robotics human behavior morality artificial intelligence

shared by Steve Bosserman on 23 Apr 18 - No Cached
  • The biggest concern is that we might one day create conscious machines: sentient beings with beliefs, desires and, most morally pressing, the capacity to suffer. Nothing seems to be stopping us from doing this. Philosophers and scientists remain uncertain about how consciousness emerges from the material world, but few doubt that it does. This suggests that the creation of conscious machines is possible.
  • If we did create conscious beings, conventional morality tells us that it would be wrong to harm them — precisely to the degree that they are conscious, and can suffer or be deprived of happiness. Just as it would be wrong to breed animals for the sake of torturing them, or to have children only to enslave them, it would be wrong to mistreat the conscious machines of the future.
  • Anything that looks and acts like the hosts on “Westworld” will appear conscious to us, whether or not we understand how consciousness emerges in physical systems. Indeed, experiments with AI and robotics have already shown how quick we are to attribute feelings to machines that look and behave like independent agents.
  • ...3 more annotations...
  • This is where actually watching “Westworld” matters. The pleasure of entertainment aside, the makers of the series have produced a powerful work of philosophy. It’s one thing to sit in a seminar and argue about what it would mean, morally, if robots were conscious. It’s quite another to witness the torments of such creatures, as portrayed by actors such as Evan Rachel Wood and Thandie Newton. You may still raise the question intellectually, but in your heart and your gut, you already know the answer.
  • But the prospect of building a place like “Westworld” is much more troubling, because the experience of harming a host isn’t merely similar to that of harming a person; it’s identical. We have no idea what repeatedly indulging such fantasies would do to us, ethically or psychologically — but there seems little reason to think that it would be good.
  • For the first time in our history, then, we run the risk of building machines that only monsters would use as they please.
Steve Bosserman

How Facebook Is Throwing Our Brains Into Overdrive - Pacific Standard - 0 views

psmag.com/...ebook-is-a-digital-narco-state

facebook neuroscience manipulation

shared by Steve Bosserman on 04 May 18 - No Cached
  • The human brain has always loved the dopamine rush of notifications, in any form; recent research indicates the unpredictable but ubiquitous updates of Gmail or Twitter carry the same neurological effect as rocking a slot machine. While Internet use is "not addictive in the same way as pharmacological substances are," as cognitive scientist Tom Stafford noted in 2013, we continually chase those unpredictable payoffs on Facebook and Instagram in ways that tend to mirror gambling or sex addictions, even if "Internet addiction" writ large currently holds an ambiguous position in the Diagnostic and Statistical Manual of Mental Disorders.
  • For products whose fundamental business proposition is harnessing attention, building those so-called "compulsion loops" isn't an accident of technology—it's the whole point. Indeed, observers have argued since Parker's "human psychology" flub last year that Facebook has not just meticulously measured, but fundamentally altered human behavior, and nascent technology ventures emboldened by Facebook's world-changing success have sought to translate the behavioral tricks that psychologist B.F. Skinner applied to the gambling kiosk to every mobile app under the sun. "When a gambler feels favored by luck, dopamine is released," Natasha Schüll, author of Addiction by Design: Machine Gambling in Las Vegas, told the Guardian in March. All Facebook managed to do was find a way to miniaturize the captivating logic of the slot machine—with no cost to the user but their time and attention.
  • While the human brain is tremendously plastic, that doesn't mean Facebook is savagely rewiring the human brain. Indeed, the Facebook users in the Cal State–Fullerton study "showed greater activation of their amygdala and striatum, brain regions that are involved in impulsive behavior," as Live Science's Tia Ghose reported at the time. Ghose continued: "But unlike in the brains of cocaine addicts, for instance, the Facebook users showed no quieting of the brain systems responsible for inhibition in the prefrontal cortex." Facebook isn't fundamentally rewiring the structure of the human brain, but its ubiquity has the same relative effect by kicking our rewards centers into overdrive.
Bill Fulkerson

Optimization is as hard as approximation - Machine Learning Research Blog - 0 views

francisbach.com/on-is-as-hard-as-approximation

machine learning optimization ML

shared by Bill Fulkerson on 18 Dec 20 - No Cached
  • Bill Fulkerson on 18 Dec 20
     
    Optimization is a key tool in machine learning, where the goal is to achieve the best possible objective function value in a minimum amount of time. Obtaining any form of global guarantees can usually be done with convex objective functions, or with special cases such as risk minimization with one-hidden over-parameterized layer neural networks (see the June post). In this post, I will consider low-dimensional problems (imagine 10 or 20), with no constraint on running time (thus get ready for some running-times that are exponential in dimension!).
1 - 20 of 166 Next › Last »
Showing 20▼ items per page