# Tag Info

72

As argued by Selvaraju et al., there are three stages of AI evolution, in all of which interpretability is helpful. In the early stages of AI development, when AI is weaker than human performance, transparency can help us build better models. It can give a better understanding of how a model works and helps us answer several key questions. For example why ...

51

tl;dr There are many valid reasons why people might fear (or better be concerned about) AI, not all involve robots and apocalyptic scenarios. To better illustrate these concerns, I'll try to split them into three categories. Conscious AI This is the type of AI that your question is referring to. A super-intelligent conscious AI that will destroy/enslave ...

21

You don't necessarily have to analyse it all. Just by having such data available you can achieve a lot in terms of surveillance, as long as you can retrieve relevant parts. A few years ago there was a Radiolab podcast, "The Eye in the Sky" (there's a full transcript on the site). The basic idea is that you have a plane circling a city 24/7, and filming what ...

17

Why do we need explainable AI? ... why we need to know "how does its intelligence work?" Because anyone with access to the equipment, enough skill, and enough time, can force the system to make a decision that is unexpected. The owner of the equipment, or 3rd parties, relying on the decision without an explanation as to why it is correct would be at a ...

15

Short term Physical accidents, e.g. due to industrial machinery, aircraft autopilot, self-driving cars. Especially in the case of unusual situations such as extreme weather or sensor failure. Typically an AI will function poorly under conditions where it has not been extensively tested. Social impacts such as reducing job availability, barriers for the ...

12

Lately with my Google searches, the AI model keeps auto filling the ending of my searches with: “...in Vietnamese” I can see how this would be annoying. I don't think Google's auto-complete algorithm and training data is publicly available. Also it changes frequently as they work to improve the service. As such, it is hard to tell what exactly is leading it ...

11

With AI technology at its current stage (or at least reasonably close to this stage), the jobs you proposed may very well be openings created by AI automation. However, sufficiently advanced AI technology--- the kind that can function as general purpose labor replacement--- will make even these jobs obsolete. This is because such an AI would be able to ...

10

I'd say the most successful are the ones so commonly used that we don't even notice them: The mail systems that automatically decipher handwritten addresses on your packages, they use machine vision and have probably been doing it since mid-90s. Algorithmic trading bots on stock markets - they handle something like 85% of all trades. Many modern CPUs use ...

10

One of the main arguments for self-driving cars is that presumably they'll get better and better at driving as the technology progresses, they have no temporal attention deficits or aggressive urges or drug habits and sense their environment 360°, all the while communicating with the other cars, which all together basically amounts to LESS DEAD PEOPLE. We ...

10

You would also want to consider physical limitations. If you are even storing 126 yottabyte of data per day, then if we look at the current theoretical densest data storage medium, DNA, at 215 petabytes per gram, we get... ${(126 * 10^{24}) \over (215 * 10^{15})} = 586046511$ grams per day 586046511 g = 586046 kg = 586 Metric Tonnes just for storage.

9

If you're a bank, hospital or any other entity that uses predictive analytics to make a decision about actions that have huge impact on people's lives, you would not make important decisions just because Gradient Boosted trees told you to do so. Firstly, because it's risky and the underlying model might be wrong and, secondly, because in some cases it is ...

9

In addition to the other answers, I would like to add to nuking cookie factory example: Machine learning AIs basically try to fulfill a goal described by humans. For example, humans create an AI running a cookie factory. The goal they implement is to sell as many cookies as possible for the highest profitable margin. Now, imagine an AI which is ...

8

My favorite scenario for harm by AI involves not high intelligence, but low intelligence. Specifically, the grey goo hypothesis. This is where a self-replicating, automated process runs amok and converts all resources into copies of itself. The point here is that the AI is not "smart" in the sense of having high intelligence or general intelligence--it ...

8

The answer is really very simple. If you have the dystopian power over all the mobile devices in the first place, you would not make them send all their data over to any "global data storage" just like that. Instead, you would have put a local AI on each device that filters, processes, categorizes and flags the important parts, sending only those parts plus ...

7

There are multiple motivations for self driving cars. Self driving cars have the potential to be much safer. Self driving cars are far more reliable than humans and can learn and have their software improved and upgraded, resulting in safer roads and far fewer accidents. More on self-driving car safety: http://bigthink.com/ideafeed/googles-self-...

7

Here are my suggestions Her, the AI part (movie spoiler): Ex Machina the AI part (movie spoiler): Eagle Eye, the AI part (movie spoiler): Big Hero 6, the AI part (movie spoiler):

7

Explainable AI is often desirable because AI (in particular, artificial neural networks) can catastrophically fail to do their intended job. More specifically, it can be hacked or attacked with adversarial examples or it can take unexpected wrong decisions whose consequences are catastrophic (for example, it can lead to the death of people). For instance, ...

6

2001 (1968) HAL 9000 is a great example of an artificial general intelligence that goes astray, where the humans don't understand the reasoning process as values dis-align. (This is a nod to Asimov in the sense of humans not understanding the implications of a logical framework. Marvin Minsky was an adviser on the film.) BladeRunner (1982) The critical ...

6

I have an example which goes in kinda the opposite direction of the public's fears, but is a very real thing, which I already see happening. It is not AI-specific, but I think it will get worse through AI. It is the problem of humans trusting the AI conclusions blindly in critical applications. We have many areas in which human experts are supposed to make ...

5

Automation Levels Most cars have some Level 1 automation, such as cruise control and various skid/flip probability reduction systems. Most high volume passenger vehicles have higher levels. Some military and private air, land, and sea equipment are already at Level 5. Level 4 requires that driving be automated during normal driving conditions, with manual ...

5

They would probably have followed the same sequence we do: be amazed at the capabilities, ask how it is done, wonder whether this is really intelligence and (or) point out our narrow the performance was, require more next time to be impressed again.

5

I would say the biggest real threat would be the unbalancing/disrupting we are already seeing. The changes of putting 90% the country out of work are real, and the results (which will be even more uneven distribution of wealth) are terrifying if you think them through.

5

This only intents to be a complement to other answers so I will not discuss to possibility of AI trying to willingly enslave humanity. But a different risk is already here. I would call it unmastered technology. I have been teached science and technology, and IMHO, AI has by itself no notion of good and evil, nor freedom. But it is built and used by human ...

4

Why are self-driving cars awesome? Safety: better awareness (due to more sensors), better reaction time, fewer distracted/injured/drunk/texting drivers on the road, etc Convenience: pick up my kids from school, park itself at the grocery store, take itself to be serviced, etc Faster transit: with increased safety, you can increase speed limits, with proper ...

4

Given all your assumptions about AI turn out to be true, we would have some kind of utopia, where no one has to work, and there is plenty of everything. Fair enough. Your other assumptions is about human nature, and that is where I'd challenge your conclusion: Just because there are computers better than humans at some task, that does not automatically take ...

4

The ultimate goal of machine learning is to bypass the developer... When we will have a "master algorithm" that can learn how to generalize any function or algorithm from examples, it can essentially replace any developer, skip the 'development" stage, going from problem directly to algorithm. We can't know when this will happen, but as we surrounded with ...

4

Will Artificial Intelligence some day become a problem to humanity after learning human behaviors and characteristics? It can be answered in both ways, I think. Yes, they may become a problem. With the increasing integration of loads of apps and smart devices in our life, almost everything defining an individual human being is digitalised. For instance, ...

4

AI production overseers - People who will command AI to build and control mines and factories. It's like a strategic game, but in the real world. Explanation: AI even with the intellect of a bee and ability to understand/execute commands in combination with appropriate technologies will be enough to create robots, which will be able to build mines and ...

4

AI Gatekeepers Their job would be to make sure they (AI) don't accidentally become our overlords. AI Tax Each robot that replaces a human worker is taxed. This line of thought was influenced by Bill Gates's recommendation on adding taxes for robots and an article on universal basic income. If the gold rush for AI puts many of us out of work (which is quite ...

4

One way to criticize the study could be to attack the data on which the study is based on. An image on a social network is not "neutral" (those are not ID photo) and certainly not images from a dating website (from which the data of the study come from). For example as a homosexual / heterosexual person you will perhaps put forward different attributes on ...

Only top voted, non community-wiki answers of a minimum length are eligible