June 23, 2019
Everyday Chaos coverage, etc.
I just posted a new page at the Everyday Chaos web site. It lists media coverage, talks, and other ways into the book.
Date: June 23rd, 2019 dw
June 23, 2019
I just posted a new page at the Everyday Chaos web site. It lists media coverage, talks, and other ways into the book.
April 16, 2019
Well, actually less. And more. Allow me to explain:
The first half of the first chapter of Everyday Chaos is now available at Medium. (An Editor’s Choice, no less!)
You can also read the first half of the chapter on how our model of models is changing at the Everyday Chaos site (Direct link: pdf).
At that site you’ll also find a fifteen minute video (Direct link: video) in which I attempt to explain why I wrote the book and what it’s about.
Or, you can just skip right to the pre-order button (Direct link: Amazon or IndieBound) :)
March 24, 2019
In 1948 when Claude Shannon was inventing information science [pdf] (and, I’d say, information itself), he took as an explanatory example a simple algorithm for predicting the element of a sentence. For example, treating each letter as equiprobable, he came up with sentences such as:
XFOML RXKHRJFFJUJ ZLPWCFWKCYJ FFJEYVKCQSGHYD QPAAMKBZAACIBZLHJQD.
If you instead use the average frequency of each letter, you instead come up with sentences that seem more language-like:
OCRO HLI RGWR NMIELWIS EU LL NBNESEBYA TH EEI ALHENHTTPA OOBTTVA NAH BRL.
At least that one has a reasonable number of vowels.
If you then consider the frequency of letters following other letters—U follows a Q far more frequently than X does—you are practically writing nonsense Latin:
ON IE ANTSOUTINYS ARE T INCTORE ST BE S DEAMY ACHIN D ILONASIVE TUCOOWE AT TEASONARE FUSO TIZIN ANDY TOBE SEACE CTISBE.
Looking not at pairs of letters but triplets Shannon got:
IN NO IST LAT WHEY CRATICT FROURE BIRS GROCID PONDENOME OF DEMONSTURES OF THE REPTAGIN IS REGOACTIONA OF CRE.
Then Shannon changes his units from triplets of letters to triplets of words, and gets:
THE HEAD AND IN FRONTAL ATTACK ON AN ENGLISH WRITER THAT THE CHARACTER OF THIS POINT IS THEREFORE ANOTHER METHOD FOR THE LETTERS THAT THE TIME OF WHO EVER TOLD THE PROBLEM FOR AN UNEXPECTED.
Pretty good! But still gibberish.
Now jump ahead seventy years and try to figure out which pieces of the following story were written by humans and which were generated by a computer:
In a shocking finding, scientist discovered a herd of unicorns living in a remote, previously unexplored valley, in the Andes Mountains. Even more surprising to the researchers was the fact that the unicorns spoke perfect English.
The scientist named the population, after their distinctive horn, Ovid’s Unicorn. These four-horned, silver-white unicorns were previously unknown to science.
Now, after almost two centuries, the mystery of what sparked this odd phenomenon is finally solved.
Dr. Jorge Pérez, an evolutionary biologist from the University of La Paz, and several companions, were exploring the Andes Mountains when they found a small valley, with no other animals or humans. Pérez noticed that the valley had what appeared to be a natural fountain, surrounded by two peaks of rock and silver snow.
Pérez and the others then ventured further into the valley. “By the time we reached the top of one peak, the water looked blue, with some crystals on top,” said Pérez.
“Pérez and his friends were astonished to see the unicorn herd”Pérez and his friends were astonished to see the unicorn herd. These creatures could be seen from the air without having to move too much to see them – they were so close they could touch their horns.
While examining these bizarre creatures the scientists discovered that the creatures also spoke some fairly regular English. Pérez stated, “We can see, for example, that they have a common ‘language,’ something like a dialect or dialectic.”
Dr. Pérez believes that the unicorns may have originated in Argentina, where the animals were believed to be descendants of a lost race of people who lived there before the arrival of humans in those parts of South America.
While their origins are still unclear, some believe that perhaps the creatures were created when a human and a unicorn met each other in a time before human civilization. According to Pérez, “In South America, such incidents seem to be quite common.”
However, Pérez also pointed out that it is likely that the only way of knowing for sure if unicorns are indeed the descendants of a lost alien race is through DNA. “But they seem to be able to communicate in English quite well, which I believe is a sign of evolution, or at least a change in social organization,” said the scientist.
The answer: The first paragraph was written by a human being. The rest was generated by a machine learning system trained on a huge body of text. You can read about it in a fascinating article (pdf of the research paper) by its creators at OpenAI. (Those creators are: Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever.)
There are two key differences between this approach and Shannon’s.
First, the new approach analyzed a very large body of documents from the Web. It ingested 45 million pages linked in Reddit comments that got more than three upvotes. After removing duplicates and some other cleanup, the data set was reduced to 8 million Web pages. That is a lot of pages. Of course the use of Reddit, or any one site, can bias the dataset. But one of the aims was to compare this new, huge, dataset to the results from existing sets of text-based data. For that reason, the developers also removed Wikipedia pages from the mix since so many existing datasets rely on those pages, which would smudge the comparisons.
(By the way, a quick google search for any page from before December 2018 mentioning both “Jorge Pérez” and “University of La Paz” turned up nothing. “The AI is constructing, not copy-pasting.”The AI is constructing, not copy-pasting.)
The second distinction from Shannon’s method: the developers used machine learning (ML) to create a neural network, rather than relying on a table of frequencies of words in triplet sequences. ML creates a far, far more complex model that can assess the probability of the next word based on the entire context of its prior uses.
The results can be astounding. While the developers freely acknowledge that the examples they feature are somewhat cherry-picked, they say:
When prompted with topics that are highly represented in the data (Brexit, Miley Cyrus, Lord of the Rings, and so on), it seems to be capable of generating reasonable samples about 50% of the time. The opposite is also true: on highly technical or esoteric types of content, the model can perform poorly.
There are obviously things to worry about as this technology advances. For example, fake news could become the Earth’s most abundant resource. For fear of its abuse, its developers are not releasing the full dataset or model weights. Good!
Nevertheless, the possibilities for research are amazing. And, perhaps most important in the longterm, one by one the human capabilities that we take as unique and distinctive are being shown to be replicable without an engine powered by a miracle.
That may be a false conclusion. Human speech does not consist simply of the utterances we make but the complex intentional and social systems in which those utterances are more than just flavored wind. But ML intends nothing and appreciates nothing. “Nothing matters to ML.”Nothing matters to ML. Nevertheless, knowing that sufficient silicon can duplicate the human miracle should shake our confidence in our species’ special place in the order of things.
(FWIW, my personal theology says that when human specialness is taken as conferring special privilege, any blow to it is a good thing. When that specialness is taken as placing special obligations on us, then at its very worst it’s a helpful illusion.)
December 12, 2018
For the past six months I’ve been a writer in residence embedded in a machine learning research group — PAIR (People + AI Research) — at the Google site in Cambridge, MA. I was recently renewed for another 6 months.
No, it’s not clear what a “writer in residence” does. So, I’ve been writing occasional posts that try to explain and contextualize some basic concepts in machine learning from the point of view of a humanities major who is deeply lacking the skills and knowledge of a computer scientist. Fortunately the developers at PAIR are very, very patient.
Here are three of the posts:
Machine Learning’s Triangle of Error: “…machine learning systems ‘think’ about fairness in terms of three interrelated factors: two ways the machine learning (ML) can go wrong, and the most basic way of adjusting the balance between these potential errors.”
Confidence Everywhere!: “… these systems are actually quite humble. It may seem counterintuitive, but we could learn from their humility.”
Hashtags and Confidence: “…in my fever dream of the future, we routinely say things like, “That celebrity relationship is going to last, 0.7 for sure!” …Expressions of confidence probably (0.8) won’t take exactly that form. But, then, a decade ago, many were dubious about the longevity of tagging…”
I also wrote about five types of fairness, which I posted about earlier: “…You appoint five respected ethicists, fairness activists, and customer advocates to figure out what gender mix of approved and denied applications would be fair. By the end of the first meeting, the five members have discovered that each of them has a different idea of what’s fair…”
I’ve also started writing an account of my attempt to write my very own machine learning program using TensorFlow.js: which lets you train a machine learning system in your browser; TensorFlow.js is a PAIR project. This project is bringing me face to face with the details of implementing even a “Hello, world”-ish ML program. (My project aims at suggesting tags for photos, based on a set of tagged images (Creative Commons-ed) from Flickr. It’s a toy, of course.)
I have bunch of other posts in the pipeline, as well as a couple of larger pieces on larger topics. Meanwhile, I’m trying to learn as much as I possibly can without becoming the most annoying person in Cambridge. But it might be too late to avoid that title…
September 14, 2018
Google PAIR (People + AI Research) has just posted my attempt to explain what fairness looks like when it’s operationalized for a machine learning system. It’s pegged around five “fairness buttons” on the new Google What-If tool, a resource for developers who want to try to figure out what factors (“features” in machine learning talk) are affecting an outcome.
Note that there are far more than five ways to operationalize fairness. The point of the article is that once we are forced to decide exactly what we’re going to count as fair — exactly enough that a machine learning system can implement it — we realize just how freaking complex fairness is. OMG. I broke my brain trying to figure out how to explain some of those ideas, and it took several Google developers (especially James Wexler) and a fine mist of vegetarian broth to restore it even incompletely. Even so, my explanations are less clear than I (or you, I’m sure) would like. But at least there’s no math in them :)
I’ll write more about this at some point, but for me the big take-away is that fairness has had value as a moral concept so far because it is vague enough to allow our intuition to guide us. Machine learning is going to force us to get very specific about it. But we are not yet adept enough at it — e.g., we don’t have a vocabulary for talking about the various varieties — plus we don’t agree about them enough to be able to navigate the shoals. It’s going to be a big mess, but something we have to work through. When we do, we’ll be better at being fair.
Now, about the fact that I am a writer-in-residence at Google. Well, yes I am, and have been for about 6 weeks. It’s a 6 month part-time experiment. My role is to try to explain some of machine learning to people who, like me, lack the technical competence to actually understand it. I’m also supposed to be reflecting in public on what the implications of machine learning might be on our ideas. I am expected to be an independent voice, an outsider on the inside.
So far, it’s been an amazing experience. I’m attached to PAIR, which has developers working on very interesting projects. They are, of course, super-smart, but they have not yet tired of me asking dumb questions that do not seem to be getting smarter over time. So, selfishly, it’s been great for me. And isn’t that all that really matters, hmmm?
May 6, 2018
Primavera De Filippi is an expert in blockchain-based tech. She is giving a ThursdAI talk on Plantoid, an event held by Harvard’s Berkman Klein Center for Internet & Society and the MIT Media Lab. Her talk is officially on operational autonomy vs. decisional autonomy, but it’s really about how weird things become when you build a computerized flower that merges AI and the blockchain. For me, a central question of her talk was: Can we have autonomous robots that have legal rights and can own and spend assets, without having to resort to conferring personhood on them the way we have with corporations?
NOTE: Live-blogging. Getting things wrong. Missing points. Omitting key information. Introducing artificial choppiness. Over-emphasizing small matters. Paraphrasing badly. Not running a spellpchecker. Mangling other people’s ideas and words. You are warned, people. |
She begins by pointing to the 3 industrial revolutions so far: Steam led to mechanized production ; Electricity led to mass production; Electronics led to automated production. The fourth — AI — is automating knowledge production.
People are increasingly moving into the digital world, and digital systems are moving back into the physical worlds, creating cyber-physical systems. E.g., the Internet of Things senses, communicates, and acts. The Internet of Smart Things learns from the data the things collect, makes inferences, and then acts. The Internet of Autonomous Things creates new legal challenges. Various actors can be held liable: manufacturer, software developer, user, and a third party. “When do we apply legal personhood to non-humans?”
With autonomous things, the user and third parties become less liable as the software developer takes on more of the liability: There can be a bug. Someone can hack into it. The rules that make inferences are inaccurate. Or a bad moral choice has led the car into an accident.
The sw developer might have created bug-free sw but its interaction with other devices might lead to unpredictability; multiple systems operating according to different rules might be incompatible; it can be hard to identify the chain of causality. So, who will be liable? The manufacturers and owners are likely to have only limited liability.
So, maybe we’ll need generalized insurance: mandatory insurance that potentially harmful devices need to subscribe to.
Or, perhaps we will provide some form of legal personhood to machines so the manufacturers can be sued for their failings. Suing a robot would be like suing a corporation. The devices would be able to own property and assets. The EU is thinking about creating this type of agenthood for AI systems. This is obviously controversial. At least a corporation has people associated with it, while the device is just a device, Primavera points out.
So, when do we apply legal personhood to non-humans? In addition to people and corporations, some countries have assigned personhood to chimpanzees (Argentina, France) and to natural resources (NZ: Whanganui river). We do this so these entities will have rights and cannot be simply exploited.
If we give legal personhood to AI-based systems, can AI have property rights over their assets and IP? If they are legally liable, they can be held responsible for their actions, and can be sued for compensation? “Maybe they should have contractual rights so they can enter into contracts. Can they be rewarded for their work? Taxed?”Maybe they should have contractual rights so they can enter into contracts. Can they be rewarded for their work? Taxed? [All of these are going to turn out to be real questions. … Wait for it …]
Limitations: “Most of the AI-based systems deployed today are more akin to slaves than corporations.” They’re not autonomous the way people are. They are owned, controlled and maintained by people or corporations. They act as agents for their operators. They have no technical means to own or transfer assets. (Primavera recommends watching the Star Trek: The Next Generation episode “The Measure of the Man” that asks, among other things, whether Data (the android) can be dismantled and whether he can resign.)
Decisional autonomy is the capacity to make a decision on your own, but it doesn’t necessarily bring what we think of as real autonomy. E.g., an AV can decide its route. For real autonomy we need operational autonomy: no one is maintaining the thing’s operation at a technical level. To take a non-random example, a blockchain runs autonomously because there is no single operator controlling. E.g., smart contracts come with a guarantee of execution. Once a contract is registered with a blockchain, no operator can stop it. This is operational autonomy.
We are getting first example of autonomous devices using blockchain. The most famous is the Samsung washing machine that can detect when the soap is empty, and makes a smart contract to order more. Autonomous cars could work with the same model; they could not be owned by anyone and collect money when someone uses them. These could be initially purchased by someone and then buy themselves off: “They’d have to be emancipated,” she says. Perhaps they and other robots can use the capital they accumulate to hire people to work for them. [Pretty interesting model for an Uber.]
She introduces Plantoid, a blockchain-based life form. “Plantoid is autonomous, self-sufficient, and can reproduce.”It’s autonomous, self-sufficient, and can reproduce. Real flowers use bees to reproduce. Plantoids use humans to collect capital for their reproduction. Their bodies are mechanical. Their spirit is an Ethereum smart contract. It collects cryptocurrency. When you feed it currency it says thank you; the Plantoid Primavera has brought, nods its flower. When it gets enough funds to reproduce itself, it triggers a smart contract that activates a call for bids to create the next version of the Plantoid. In the “mating phase” it looks for a human to create the new version. People vote with micro-donations. Then it identifies a winner and hires that human to create the new one.
There are many Plantoids in the world. Each has its own “DNA”. New artists can add to it. E.g., each artist has to decide on its governance, such as whether it will donate some funds to charity. The aim is to make it more attractive to be contributed to. The most fit get the most money and reproduces themselves. BurningMan this summer is going to feature this.
Every time one reproduces, a small cut is given to the pattern that generated it, and some to the new designer. This flips copyright on its head: the artist has an incentive to make her design more visible and accessible and attractive.
So, why provide legal personhood to autonomous devices? We want them to be able to own their own assets, to assume contractual rights, and legal capacity so they can sue and be sued, and limit their liability. “ Blockchain lets us do that without having to declare the robot to be a legal person.” Blockchain lets us do that without having to declare the robot to be a legal person.
The plant effectively owns the cryptofunds. The law cannot affect this. Smart contracts are enforced by code
Who are the parties to the contract? The original author and new artist? The master agreement? Who can sue who in case of a breach? We don’t know how to answer these questions yet.
Can a plantoid sure for breach of contract? Not if the legal system doesn’t recognize them as legal persons. So who is liable if the plant hurts someone? Can we provide a mechanism for this without conferring personhood? “How do you enforce the law against autonomous agents that cannot be stopped and whose property cannot be seized?”
Could you do this with live plants? People would bioengineer them…
A: Yes. Plantoid has already been forked this way. There’s an idea for a forest offering trees to be cut down, with the compensation going to the forest which might eventually buy more land to expand itself.
My interest in this grew out of my interest in decentralized organizations. This enables a project to be an entity that assumes liability for its actions, and to reproduce itself.
Q: [me] Do you own this plantoid?
A: Hmm. I own the physical instantiation but not the code or the smart contract. If this one broke, I could make a new one that connects to the same smart contract. If someone gets hurt because it falls on the, I’m probably liable. If the smart contract is funding terrorism, I’m not the owner of that contract. The physical object is doing nothing but reacting to donations.
Q: But the aim of its reactions is to attract more money…
A: It will be up to the judge.
Q: What are the most likely senarios for the development of these weird objects?
A: A blockchain can provide the interface for humans interacting with each other without needing a legal entity, such as Uber, to centralize control. But you need people to decide to do this. The question is how these entities change the structure of the organization.
April 27, 2018
Ben Green is giving a ThursdAI talk on “The Limits, Perils, and Challenges of ‘Fair’ Algorithms for Criminal Justice Reform.”
NOTE: Live-blogging. Getting things wrong. Missing points. Omitting key information. Introducing artificial choppiness. Over-emphasizing small matters. Paraphrasing badly. Not running a spellpchecker. Mangling other people’s ideas and words. You are warned, people. |
In 2016, the COMPAS algorithm
became a household name (in some households) when ProPublica showed that it predicted that black men were twice as likely as white men to jump bail. People justifiably got worried that algorithms can be highly biased. At the same time, we think that algorithms may be smarter than humans, Ben says. These have been the poles of the discussion. Optimists think that we can limit the bias to take advantage of the added smartness.
There have been movements to go toward risk assessments for bail, rather than using money bail. E.g., Rand Paul and Kamala Harris have introduced the Pretrial Integrity and Safety Act of 2017. There have also been movements to use scores only to reduce risk assessments, not to increase them.
But are we asking the right questions? Yes, the criminal justice system would be better if judges could make more accurate and unbiased predictions, but it’s not clear that machine learning can do this. So, two questions: 1. Is ML an appropriate tool for this. 2. Is implementing MK algorithms an effective strategy for criminal justice reform?
#1 Is ML and appropriate tool to help judges make more accurate and unbiased predictions?
ML relies on data about the world. This can produce tunnel vision by causing us to focus on particular variables that we have quantified, and ignore others. E.g., when it comes to sentencing, a judge balances deterrence, rehabilitation, retribution, and incapacitating a criminal. COMPAS predicts recidivism, but none of the other factors. This emphasizes incapacitation as the goal of sentencing. This might be good or bad, but the ML has shifted the balance of factors, framing the decision without policy review or public discussion.
Q: Is this for sentencing or bail? Because incapacitation is a more important goal in sentencing than in bail.
A: This is about sentencing. I’ll be referring to both.
Data is always about the past, Ben continues. ML finds statistical correlations among inputs and outputs. It applies those correlations to the new inputs. This assumes that those correlations will hold in the future; it assumes that the future will look like the past. But if we’re trying reform the judicial system, we don’t want the future to look like the past. ML can thus entrench historical discrimination.
Arguments about the fairness of COMPAS are often based on competing mathematical definitions of fairness. But we could also think about the scope of what we couint as fair. ML tries to make a very specific decision: among a population, who recidivates? If you take a step back and consider the broader context of the data and the people, you would recognize that blacks recidivate at a higher rate than whites because of policing practices, economic factors, racism, etc. Without these considerations, you’re throwing away the context and accepting the current correlations as the ground truth. Even if we were to change the base data, the algorithm wouldn’t make the change, unless you retrain it.
Q: Who retrains the data?
A: It depends on the contract the court system has.
Algorithms are not themselves a natural outcome of the world. Subjective decisions go into making them: which data to input, choosing what to predict, etc. The algorithms are brought into court as if they were facts. Their subjectivity is out of the frame. A human expert would be subject to cross examination. We should be thinking of algorithms that way. Cross examination might include asking how accurate the system is for the particular group the defendant is in, etc.
Q: These tools are used in setting bail or a sentence, i.e., before or after a trial. There may not be a venue for cross examination.
In the Loomis case, an expert witness testified that the algorithm was misused. That’s not exactly what I’m suggesting; they couldn’t get to all of it because of the trade secrecy of the algorithms.
Back to the framing question. If you can make the individual decision points fair we sometimes think we’ve made the system fair. But technocratic solutions tend to sanitize rather than alter. You’re conceding the overall framework of the system, overlooking more meaningful changes. E.g., in NY, 71% of voters support ending pre-trial jail for misdemeanors and non-violent felonies. Maybe we should consider that. Or, consider that cutting food stamps has been shown to increases recidivism. Or perhaps we should be reconsidering the wisdom of preventative detention, which was only introduced in the 1980s. Focusing on the tech de-focuses on these sorts of reforms.
Also, technocratic reforms are subject to political capture. E.g., NJ replaced money bail with a risk assessment tool. After some of the people released committed crimes, they changed the tool so that certain crimes were removed from bail. What is an acceptable risk level? How to set the number? Once it’s set, how is it changed?
Q: [me] So, is your idea that these ML tools drive out meaningful change, so we ought not to use them?
A: Roughly, yes.
[Much interesting discussion which I have not captured. E.g., Algorithms can take away the political impetus to restore bail as simply a method to prevent flight. But sentencing software is different, and better algorithms might help, especially if the algorithms are recommending sentences but not imposing them. And much more.]
2. Do algorithms actually help?
How do judges use algorithms to make a decision? Even if the algorithm were perfect, would it improve the decisions judges make? We don’t have much of an empirical answer.
Ben was talking to Jeremy Heffner at Hunch Lab. They make predictive policing software and are well aware of the problem of bias. (“If theres any bias in the system it’s because of the crime data. That’s what we’re trying to address.” — Heffner) But all of the suggestions they give to police officers are called “missions,” which is in the military/jeopardy frame.
People are bad at incorporating quantitative data into decisions. And they filter info through their biases. E.g., the “ban the box” campaign to remove the tick box about criminal backgrounds on job applications actually increased racial discrimination because employers assumed the white applicants were less likely to have arrest records. (Agan and Starr 2016) Also, people have been shown to interpret police camera footage according to their own prior opinions about the police. (sommers 2016)
Evidence from Kentucky (Stevenson 2018): after mandatory risk assessments for bail only made a small increase in pretrial release, and these changes eroded over time as judges returned to their previous habits.
So, we need to be asking the empirical question of how judges actual use these decisions. And should judges incorporate these predictions into their decisions?
Ben’s been looking at the first question:L how do judges use algorithmic predictions? He’s running experiments on Mechanical Turk showing people profiles of defendants — a couple of sentences about the crime, race, previous record arrest record. The Turkers have to give a prediction of recidivism. Ben knows which ones actually recidivated. Some are also given a recommendation based on an algorithmic assessment. That risk score might be the actual one, random, or biased; the Turkers don’t know that about the score.
Q: It might be different if you gave this test to judges.
A: Yes, that’s a limitation.
Q: You ought to give some a percentage of something unrelated, e.g., it will rain, just to see if the number is anchoring people.
A: Good idea
Q: [me] Suppose you find that the Turkers’ assessment of risk is more racially biased than the algorithm…
A: Could be.
[More discussion until we ran out of time. Very interesting.]
April 5, 2018
I’m at a ThursdAI talk (Harvard’s Berkman Klein Center for Internet & Society and MIT Media Lab) being given by Neil Gaikwad (Twitter: @neilthemathguy, a Ph.D. at the MediaLab, in the Space Enabled Group.
NOTE: Live-blogging. Getting things wrong. Missing points. Omitting key information. Introducing artificial choppiness. Over-emphasizing small matters. Paraphrasing badly. Not running a spellpchecker. Mangling other people’s ideas and words. You are warned, people. |
Markets and institutions are parts of complex ecosystem, Neil says. His research looks at data from satellites that show how the Earth is changing: crops, water, etc. Once you’ve gathered the data, you can use machine learning to visualize the changes. There are ecosystems, including of human behavior, that are affected by this. It affects markets and institutions. E.g., a drought may require an institutional response, and affect markets.
Traditional markets, financial markets, and gig economies all share characteristics. Farmers markets are complex ecosystems of people with differing information and different amounts of it, i.e. asymmetric info. Same for financial markets. Same for gig economies.
Indian markets have been failing; there have been 300,000 suicides in the last 30 years. Stock markets have crashed suddenly due to blackbox marketing; in some cases we still don’t know why. And London has banned Uber. So, it doesn’t matter which markets or institutions we look at, they’re losing our trust.
An article in New Scientist asked what we can do to regain this trust. For black box AI, there are questions of fairness and equity. But what would human-machine collaboration be like? Are there design principles for markets.?
Neil stops for us to discuss.
Q: How do you define the justice?
A: Good question. Fairness? Freedom? The designer has a choice about how to define it.
Q: A UN project created an IT platform that put together farmers and direct consumers. The pricing seemed fairer to both parties. So, maybe avoid intermediaries, as a design principle?
Neil continues. So, what is the concept of justice here?
1. Rawls and Kant: Transcendental institutionalism. It’s deontological: follow a principle for perfect justice. Use those principles to define a perfect institution. The properties are defined by a social contract. But it doesn’t work, as in the examples we just saw. What is missing. People and society. [I.e., you run the institution according to principles, but that doesn’t guarantee that the outcome will be fair and just. My example: Early Web enthusiasts like me thought the Web was an institution built on openness, equality, creative anarchy, etc., yet that obviously doesn’t ensure that the outcome will share those properties.]
2. Realized-focused institutionalism (Sen
2009): How to reverse this trend. It is consequentialist: what will be the consequences of the design of an institution. It’s a comparative assessment of different forms of institutions. Instead of asking for the perfectly justice society, Sen asks how justice can be advanced. The most critical tool for evaluating any institution is to look at how it actually realizes how people’s lives change.
Sen argues that principles are important. They can be expressed by “niti,” Sanskrit for rules and institutions. But you also need nyaya: a form of social arrangement that makes sure that those rules are obeyed. These rules come from social choice, not social contract.
Example: Gig economies. The data comes from mechanical turk, upwork, crowdflower, etc. This creates employment for many people, but it’s tough. E.g., identifying images. Use supervised learning for this. The Turkers, etc., do the labelling to train the image recognition system. The Turkers make almost no money at this. This is the wicked problem of market design: The worker can have identifications rejected, sometimes with demeaning comments.
“The Market for Lemons” (Akerlog, et al., 1970): all the cars started to look alike and now all gig-workers look alike to those who hire them: there’s no value given to bringing one’s value to the labor.
So, who owns the data? Who has a stake in the models? In the intellectual property?
If you’re a gig worker, you’re working with strangers. You don’t know the reputation of the person giving me data. Or renting me the Airbnb apartment. So, let’s put a rule: reputation is the backbone. In sharing economies, most of the ratings are the highest. Reputation inflation. So, can we trust reputation? This happens because people have no incentive to rate. There’s social pressure to give a positive rating.
So, thinking about Sen, can we think about an incentive for honest reputation? Neil’s group has been thinking about a system [I thought he said Boomerang, but I can’t find that]. It looks at the workers’ incentives. It looks at the workers’ ratings of each other. If you’re a requester, you’ll see the workers you like first.
Does this help AI design?
MoralMachine has had 1.3M voters and 18M pairwise comparisons (i.e., people deciding to go straight or right). Can this be used as a voting based system for ethical decision making (AAAI 2018)? You collect the pairwise preferences, learn the model of preference, come to a collective preference, and have voting rules for collective decision.
Q: Aren’t you collect preferences, not normative judgments? The data says people would rather kill fat people than skinny ones.
A: You need the social behavior but also rules. For this you have to bring people into the loop.
Q: How do we differentiate between what we say we want and what we really want?
A: There are techniques, such as “Bayesian Truth Serum”nomics.mit.edu/files/1966”>Bayesian Truth Serum.
Conclusion: The success of markets, institutions or algorithms, is highly dependent on how this actually affects people’s lives. This thinking should be central to the design and engineering of socio-technical systems.
April 2, 2018
“If a lion could talk, we could not understand him.”
— Ludwig Wittgenstein, Philosophical Investigations, 1953.
“If an algorithm could talk, we could not understand it.”
— Deep learning, Now.
March 19, 2018
Kate Zwaard (twitter: @kzwa) Chief of National Digital Strategies at the Library of Congress and leader of the LC Lab, is opening MIT Libraries’ Grand Challenge Summit..The next 1.5 days will be about the grand challenges in enabling scholarly discovery.
NOTE: Live-blogging. Getting things wrong. Missing points. Omitting key information. Introducing artificial choppiness. Over-emphasizing small matters. Paraphrasing badly. Not running a spellpchecker. Mangling other people’s ideas and words. You are warned, people. |
For context she tells us that the LC is the largest library in the world, with 164M items. It has the world’s largest collection of film, maps, comic books, telephone directories, and more. [Too many for me to keep this post up with.]
You can wolk for two football fields just in the maps section. The world’s largest collection of recorded sound. The largest collection
Personal papers from Ben Franklin, Rosa Parks, Groucho Marx, Claude Shannon, and so many more.
Last year they circulated almost a million physical items.
Every week 11,000 tangible items come in through the Copyright office.
Last year, they digitized 4.7M iems, as well 730M documents crawled from the Web, plus much more. File count: 243M and growing every day.
These serve just one of the LC’s goal: “Acquire, preserve, and provide access to a universal collection of knowledge and the record of America’s creativity.” Not to mention serving Congress, and much more. [I can only keep up with a little of this. Kate’s a fantastic presenter and is not speaking too quickly. The LC is just too big!]
Kate thinks of the LC’s work as an exothermic reaction that needs an activation energy or catalyst. She leads the LC Labs, which started a year ago as a place of experimentation. The LC is a delicate machine, which makes it hard for it to change. The Labs enable experimentation. “Trying things that are easy and cheap is the only way forward.”
When thinking about what to do next, she things about what’s feasible and the impact. One way of having impact: demonstrating that the collection has unexplored potentials for research. She’s especially interested in how the Labs can help deal with the problem of scale at the LC.
She talks about some of Lab’s projects.
If you wanted to make stuff with LC data, there was no way of doing that. Now there’s LC for Robots, added documentation, and Jupyter Notebooks: an open source Web app that let you create open docs that contain code, running text, etc. It lets people play with the API without doing all the work from scratch.
But it’s not enough to throw some resources onto a Web page. The NEH data challenge asked people to create new things using the info about 12M newspapers in the collection. Now the Lab has the Congressional Data Challenge: do something with with Congressional data.
Labs has an Innovator in Residence project. The initial applicants came from LC to give it a try. One of them created a “Beyond Words” crowdsourcing project that asks them to add data to resources
Kate likes helping people find collections they otherwise would have missed. For ten years LC has collaborated wi the Flickr Commons. But they wanted to crowdsource a transcription project for any image of text. A repo will be going up on GitHub shortly for this.
In the second year of the Innovator in Residence, they got the artist Jer Thorp [Twitter: @blprnt] to come for 6 months. Kate talks about his work with the papers of Edward Lorenz, who coined the phrase “The Butterfly Effect.” Jer animated Lorenz’s attractor, which, he points out, looks a bit like a butterfly. Jer’s used the attractor on a collection of 3M words. It results in “something like a poem.” (Here’s Jer’s Artist in the Archive podcast about his residency.)
Jer wonders how we can put serendipity back into the LC and into the Web. “How do we enable our users to be carried off by curiousity not by a particular destination.” The LC is a closed stack library, but it can help guide digital wanderers. ”
Last year the LC released 25M catalog records. Jer did a project that randomly pulls the first names of 20 authors in any particular need. It demonstrates, among other things, the changing demographics of authors. Another project: “Birthy Deathy” that displays birthplace info. Antother looks for polymaths.
In 2018 the Lab will have their first open call for an Innovator in Residence. They’ll be looking for data journalists.
Kate talks about Laura Wrubel
‘s work with the Lab. “Library of Congress Colors” displays a graphic of the dominant colors in a collection.
Or Laura’s Photo Roulette: you guess the date of a photo.
Kate says she likes to think that libraries not just “book holes.” One project: find links among items in the archives. But the WARC format is not amenable to that.
The Lab is partnering with lots of great grops, including JSONstor and WikiData.
They’re working on using machine learning to identify place names in their photos.
What does this have to do with scale, she asks, nothng that the LC has done pretty well with scale. E.g., for the past seven years, the size of their digital collection has doubled every 32 months.
The Library also thinks about how to become a place of warmth and welcome. (She gives a shout out to MIT Libraries’ Future of Libraries
report). Right now, visitors and scholars go to different parts of the building. Visitors to the building see a monument to knowledge, but not a living, breathing place. “The Library is for you. It is a place you own. It is a home.”
She reads from a story by Ann Lamott.
How friendship relates to scale. “Everything good that has happened in my life has happened because of friendship.” The average length of employment of a current employee is thirty years. — that’s not the average retirement year. “It’s not just for the LC but for our field.” Good advice she got: “Pick your career by the kind of people you like to be around.” Librarians!
“We’ve got a tough road ahead of us. We’re still in the early days of the disruption that computation is going to bring to our profession.” “Friendship is what will get us through these hard times. We need to invite peopld into the tent.” “Everything we’ve accomplished has been through the generosity of our friends and colleagues.” This 100% true of the Labs. It’s ust 4 people, but everything they do is done in collaboration.
She concludes (paraphrasing badly): I don’t believe in geniuses, and i don’t believe in paradigm shirts. I believe in friendship and working together over the long term. [She put this far better.]
Q&A
Q: How does the Lab decide on projects?
A: Collaboratively
Q: I’m an archivist at MIT. The works are in closed stack, which can mislead people about the scale. How do we explain the scale in an interesting way.
A: Funding is difficult because so much of the money that comes is to maintain and grow the collection and services. It can be a challenge to carve out funding for experimentation and innovation. We’ve been working hard on finding ways to help people wrap their heads around the place.
Q: Data science students are eager to engage, e.g., as interns. How can academic institutions help to make that happen?
A: We’re very interested in what sorts of partnerships we can create to bring students in. The data is so rich, and the place is so interesting.
Q: Moving from models that think about data as packages as opposed to unpacking and integrating. What do you think about the FAIR principle: making things Findable, Accesible Interoperable, and Reusable? Also, we need to bring in professionals thinking about knowledge much more broadly.
I’m very interested in Hathi Trust‘s data capsules. Are there ways we can allow people to search through audio files that are not going to age into the commons until we’re gone? You’re right: the model of chunks coming in and out is not going to work for us.
Q: In academia, our focus has been to provide resources efficiently. How can weave in serendipity without hurting the efficiency?
A: That’s hard. Maybe we should just serve the person who has a specific purpose. You could give ancillary answers. And crowdsourcing could make a lot more available.
[Great talk.]