This article is weird. I work with ML (AI is overloaded term) and don't recognize myself in this article at all. It seems to be written for managers, politicians, or economists or something.
It is like stating: "The function that gives software value is the ability to create if-then statements." Both remotely true and meaningless.
Conflating analysis with predictive modeling, pretending self-driving cars are a thing of the last decade (and not fully operational since the 80s) and this:
> “So what’s going to happen is that these prediction machines are going to make predictions better and faster and cheaper, and when you do that, two things happen. The first is that we will do a lot more predicting. And the second is that we will think of new ways of doing things for problems where the missing bit was prediction.”
If using ML or DL qualifies as a subset of AI, then AI qualifies as a subset of software and IT. Turning above statement into:
> “So what’s going to happen is that these computer are going to run code better and faster and cheaper, and when you do that, two things happen. The first is that we will do a lot more coding. And the second is that we will think of new ways of doing things for problems where the missing bit was software.”
Then you are still correct, it is a safe bet, but you are correct about a very insignificant thing.
Glad I am not the only one. This seems like a task of memory, not author identification. How could this be used at test time?
10.000 users is meaningless on a social network with millions of accounts.
What about the static features (like account creation dates)? Aren't those overfitting with cross-validation? Would not learning curves be required when classifying on unseen future data (the reason d'etre of ML)?
I firmly believe (though in these matters it is hard to prove) that China and the US are in a technological AI war, with the first to largely automate their economy getting the biggest piece of the pie. One needs pervasive surveillance to accomplish economy automation.
I believe stories like these carry a certain propaganda element and are directed behind the scenes. What ultimately sticks is "Don't share your tech with the Chinese or they will use it to build 1984's Orwell. Go work for harmless Silicon Valley instead, so you can make people click on advertisements and get them addicted to your platform.".
If we had access to all the facts (we don't), we could honestly compare the US's surveillance apparatus to the Chinese. I believe that it was the US that started it with the early Echelon systems, forcing the Chinese to step up their game (every time a Chinese spy got caught with these systems, their picture of US capabilities got a bit more clear).
All modern ML is build on old military projects, adversarial images is researched due to the military wanting no mistakes, and all popular tools see investments by DARPA/IARPA. There is just no way to escape the military when working with advanced technology, except for putting on the blinders and pretend that your work/code/tutorials are not being consumed by (foreign) intelligence agencies.
The US government is legit afraid to lose to China, because China seems to care way less about the unfairness and biases presenting itself in IT systems, while the educated US citizens demand fair automated treatment and justification. In the eyes of progress, those are mere hindrances and roadblocks that need clearing first, giving China a head start. The only thing you can do to lessen this drawback, is to publish wide and far that China does not care about ethics in AI, turning it into a PR problem for them.
DeepMind winning at Go would be like Alibaba winning the Superbowl with robots. It was a huge wake up call, and I think it rattled some cages of foreign militaries.
I myself share a lot of information with the US, including KYC data. A lot of US companies try to track every move I make online. Even if I wanted to go 1 month without touching anything Akamai, I could not. Commercialized mass surveillance it not much better than state-led mass surveillance.
It would have been interesting to see how the US would be portrayed if it was not the top dog. Like the US media attacks and publicizes the poor rights of women in countries like Afghanistan as part of the war effort, I wonder what aspects of US culture a country like Afghanistan would attack/deem subhuman.
> "Suits make a corporate comeback," says the New York Times. Why does this sound familiar? Maybe because the suit was also back in February, September 2004, June 2004, March 2004, September 2003, November 2002, April 2002, and February 2002.
> Why do the media keep running stories saying suits are back? Because PR firms tell them to. One of the most surprising things I discovered during my brief business career was the existence of the PR industry, lurking like a huge, quiet submarine beneath the news. Of the stories you read in traditional media that aren't about politics, crimes, or disasters, more than half probably come from PR firms.
> I know because I spent years hunting such "press hits." Our startup spent its entire marketing budget on PR: at a time when we were assembling our own computers to save money, we were paying a PR firm $16,000 a month. And they were worth it. PR is the news equivalent of search engine optimization; instead of buying ads, which readers ignore, you get yourself inserted directly into the stories.
I sometimes think I am like a storm glass or barometer. If there is a huge storm brewing that meter is going up or down. From my perspective I am doing exactly what I am thinking about doing, this storm really wants to make me go up or down. Of course, I have a limited view on reality, I may not even notice the storm itself.
But cognitive science, only through study of brain lesions and experiments, can offer glimpses of what is out there. What the weather really is like.
But what if the very act of categorization was an error to begin with? Causal inference poses problems like: Does the barometer change cause the storm, or does the storm cause the barometer change? These can be better solved by saying: The pressure in the barometer changing _is_ (part of) the storm. Instead of saying: If I go up, I cause the storm to follow ("If I am thinking I am a single agent, my consciousness must be singular").
In the end you are free, and I encourage you to, call it a simplified model, not an illusion. But to discard all of Dennett's consciousness philosophy on the basis of a poorly chosen word, is not a valid or fruitful conclusion. You'll miss the memetic good sauce that cures Naive Realism: https://en.wikipedia.org/wiki/Na%C3%AFve_realism
I myself, personally, prefer RAW's Maybe Logic approach to consciousness, though that it arguably less academically sound (though not less wise for it): https://www.youtube.com/watch?v=A7N6TOFyrLg
> But this illusion is good enough to allow us to act in the world. We don't call it illusion, we just call it the world.
I like calling it a "world model". Optical illusions proof that this "world model" can be consistently tricked for a wide range of humans. Now the "illusion of consciousness" is a categorization error: You apply your "world model" to your own internal processes / consciousness. So far so good. But this does not give you the right to claim that consciousness is your "world model" view of it. If it was, then so could people who took LSD and had their "world model" believe they could swim in the sky, claim to change reality/ontology/the world for all of us.
The illusion is that your world model of consciousness does not equal consciousness in reality, as proven by science, despite how clearly it may appear to you (not that conscious experience itself is an illusion).
> Most of the time I act as a single intelligent agent, not as a bunch of subsystems with no unifying goals.
This is the illusion. To you it seems that you act as a single intelligent agent, unaware of the thousands of majority votes by your senses/neurons that lead up to that point.
Both your view as a single acting agent and this illusion are still legit. But again, it is a categorization error to conclude that your experience validates that humans are all individually acting single intelligent agents. That's Descartes-era philosophy and the legacy that Dennett was railing against: "I think therefor I am" becomes "I think that I am a single agent, therefor I am such".
You can compare "hunger" with "consciousness". Disease can make one feel not hungry, while medical investigation shows that the body is in desperate need of sustenance, or vice versa. Now is it an illusion/delusion to say: "I am hungry" when your body is already full? Maybe. But it becomes a mistake when you declare "I feel hungry, therefor my body needs more sustenance". That seemingly logical conclusion is the result of an illusion.
"The line we placed there was arbitrary. The division it creates is the very cause of the problems of consciousness. Science experiments show that it is likely at the wrong location, no matter how confident we were when we drew it. I will attempt to solve these artificial problems of consciousness by showing that it is a miscategorization and not-well-justified to draw the line in the first place."
I think it was because in a time where objectivism ruled, consciousness was still a highly subjective subject. "Consciousness Explained" was an effort to discover the other side of the coin.
A subjectivist sees art, and reasons that the art object is fully formed inside his/her mind. An objectivist sees art, and reasons that the art object is entirely contained in the physical manifestation of it. Consciousness Explained was an attempt to show this objectivist view of consciousness.
Later on he tried to marry both in heterophenomonology: "The sky seems blue to non-colorblind subjects, but objectively, in reality, it part of the human-visible spectrum of wave-length n.". This framework still gives legitimacy to the subjective experience of a colorblind person, without allowing this to change physical reality (for consciousness: allowing for personal experience and realization of it, without allowing this to change the neuroscience/ontology "i experience sequential thought, therefor thoughts must be sequential, not parallel.").
It addresses the issue by telling you: Don't worry about consciousness in the first place. How your higher-level cognitive faculties see and categorize consciousness is proven to be an illusion. Eliminate this singular consciousness, this artificial construct, and any hard or soft problem disappears with it, and we can start defining conscious experience proper, instead of trying the impossible and give a solution to everyone's subjective (and possibly eternally conflicting) theatric view of consciousness.
It is not always the goal (or possible) to create a complete model of conscious experience that is indistinguishable from reality, yet one may model the path of a hurricane without getting blown away by the wind.
When people imagine their consciousness/free will faculty as its own object, they sometimes see a theatre/stage, where they are in the audience and can observe the central point where consciousness ends up forming, or a deamon sitting in front of a bunch of levers and pulling one to make a conscious free will decision.
Yet, Dennett shows that there is no such single centralized point where consciousness emerges, just a brain creating that illusion to make a higher-level lossy sense of things (consequence of modeling your own cognitive processes using Descartes-era philosophy is that you have to place an artificial cut-off somewhere). Cognitive science has experiments like the Stroop Test [1], which proof that different cognitive faculties operate at different speeds and in parallel. They may even compete for attention. Consciousness is thus distributed and a "multi-agent" system, you may not even become "consciously" aware of what the fast operating faculties are processing and feeding up to more complex faculties. Other cognitive science reaction speed experiments have shown that data is processed (helping you make decisions) before it enters conscious thought (so all "behind the stage").
No, and some philosophies deem consciousness unnecessary/irrelevant. To them, consciousness is a byproduct of brain activity, much like the heat generated from a burning lamp.
Let's say you add these digital signature variables to your credit risk scoring model anyway. The model then falls prey to confusing correlation with causation. What happens to the performance of the model?
You still owe any debts after 7 years (up to 15 years in some states). Zombie debt is not "completely clear". But ok, read "mess up your credit score for 7 years" instead. Then point remains: Marginal high-risk credit underwriting is not only dangerous to the institution, but also to the receivers of the loans (and the economy in general). It is socially perverse to hook lower-income people to consumer credit. To have middle-income people lose their house.
People with the exact same FICO score have different default rates, if you manage to bin them by race: Asians > Caucasians > Latin > Black.
> We do get a some variables even if we are not allowed to use them.
Cool, so you had access to an ethnicity variable to measure its proxy power and significance? I feel this is important and very rare outside of Europe.
Let's change the game. You can allocate an investment to either of two banks. When building credit scoring models, one bank has access to just FICO scores, the other bank also has access to FICO scores in addition to behavioral and signature data. Which bank do you allocate your cash to?
Now change the game so FICO is unavailable: For instance, when micro-lending to third-world country entrepreneurs. Do you still feel these digital signatures are irrelevant to making better credit risk decisions?
Let's play a game. You are in charge of a large pile of cash and want to make it grow by giving loans. Each day, two people apply, and you can give out one loan (you will have to rank the applicants). When people de-fraud you, you lose all of the loan. When people don't or can't pay you back, you lose all of the loan. When people pay back the loan, you make a little money.
Day1: User Agent: iPhone latest vs. Windows XP
Day2: Referral: Facebook friend vs. search "cheapest loans"
Day3: Time of interaction: 21:30 vs. 04:30
Day4: Email: [email protected] vs. [email protected]
Day5: Funnel: Someone who spend 10 seconds vs. someone who spend 10 minutes, made a mistake in the name, entered an email address, then deleted it, and entered another email address at a different provider.
Now if your feeling does not point you to the first applicant every day, you look at the data for guidance. You find that the number of fraudsters and non-pay's is statistically significantly higher for people with the second set of characteristics.
The alternative is to use third-party data providers. That's another can of worms. Or flip a coin and start gambling proper.
You can overdo fairness and cause troubles to the poor. Very concretely: Giving someone a loan, despite their credit score being marginal, will severely mess up their credit score forever, if they can't repay you.
If you are poor, then don't create more debt! It should be hard to rack up such a debt, not easy and accessible. No amount of credit is going to increase your social status, because you have to pay it back with your own current/near-future money.
If we want social justice for the poor through access to more money, then capitalism is not a good way to go. The state should become a credit provider.
DARPA/IARPA now invests in search 2.0: Instead of asking it to retrieve stored information, you can instruct an agent/search bot to perform tasks for you.
For instance, one should be able to search for: "who is the leader of this IRC hacker group?" "where can heroine be bought on the deep web?" "Is there women trafficking going on behind that log-in wall?" and then an intelligent agent is dispatched, avoiding/crossing roadblocks, like log-in forms, and will eventually bring you the answer.
Other possible future research areas in information retrieval include being able to search for services ("Where is cheapest taxi service for current location?") and an integration with IOT.
It is like stating: "The function that gives software value is the ability to create if-then statements." Both remotely true and meaningless.
Conflating analysis with predictive modeling, pretending self-driving cars are a thing of the last decade (and not fully operational since the 80s) and this:
> “So what’s going to happen is that these prediction machines are going to make predictions better and faster and cheaper, and when you do that, two things happen. The first is that we will do a lot more predicting. And the second is that we will think of new ways of doing things for problems where the missing bit was prediction.”
If using ML or DL qualifies as a subset of AI, then AI qualifies as a subset of software and IT. Turning above statement into:
> “So what’s going to happen is that these computer are going to run code better and faster and cheaper, and when you do that, two things happen. The first is that we will do a lot more coding. And the second is that we will think of new ways of doing things for problems where the missing bit was software.”
Then you are still correct, it is a safe bet, but you are correct about a very insignificant thing.