equality

Wake Up Call - Data Bias and Corporations

I don’t think some of you are going to like this blog post. I reckon my pragmatism on this matter will be thrown to the wolves of optimism and idealism, but I understand that. In fact I welcome it. This post is not meant in any way to encourage or even justify this behavior. Rather it is meant to highlight that certain characters in this equation do not have their goals aligned with the my crowd, the AI Safety crowd, regardless of their flowery rhetoric.

Corporations exist for one reason, the benefit of their shareholders. They do not exist to make your life better, or easier. They do not exist to employ people. They do not exist to pay taxes or provide a social good. They exist to maximize profits for their shareholders.

Now, sometimes companies will identify social good, community employment, bettering life and some of the other peripherals as goals of their company, but these are secondary goals. To be achieved, if the cost to the bottom line is sufficiently low. Let me illustrate. Company A makes $1 million and says that they want to employ people in their community. Every time their hire someone it costs them $100k to do so. Will they hire 11 people, no way. Certainly they won’t hire 10, probably not even 9. Instead they will employ the right amount of people to help $1 mil be maintained and grown. Depending upon their outlook on the market, that might mean hiring zero people or 100, but it is dependent upon their analysis of profit ONLY. Their hiring decision is secondary, a function of the profit math and it does not occur because of their ancillary hiring goal. Any attempts to convince the company of the values of social good may be factored into their analysis of how can they maintain and grow that $1 mil profit. If that social good has a small probability to erode that $1 mil profit then it will be pushed aside. This is not a debate, there is no “yeah but”. Corporations have a long and distinguished track record of seeking profit first. It’s what they were designed for.

With that background let’s talk about data bias. For those unfamiliar, let me define data bias:

Bias — Oxford definitions

NOUN

  1. Inclination or prejudice for or against one person or group, especially in a way considered to be unfair.
  2. A concentration on or interest in one particular area or subject.

3. A systematic distortion of a statistical result due to a factor not allowed for in its derivation.

VERB

  1. Cause to feel or show inclination or prejudice for or against someone or something.

With the increased use of artificial intelligence (AI) and machine learning (ML), companies acquire enormous amounts of data. The data is the input/fuel for AI and ML to learn and inform their decision making process. This data when acquired from a myriad of sources can acquire bias, thus the term data bias. These biases may appear as a result of the collection process, the organization of, the interpretation of or the implementation of the data. OR, the bias may be embedded in the data from the source itself such as society. There is plenty of bias in society today and thus there is plenty of bias in our data sets.

So it fairly safe to assume that much if not all of the data being input into these systems will have some amount of bias to them. AI safety advocates like myself claim that companies should monitor for and remove bias. But here is the problem. This endeavor is likely to result in sub optimal results to their AI and ML. Let me explain.

If you are trying to model and analyze the best design for your widget, you will use your ML and AI to help you identify the key criteria for that product to create demand or meet demand in the marketplace. The goal is to maximize sales of your widget. If the company takes in data, manipulates ita and during that process causes data bias to be introduced into a reasonably pure data set — they likely corrupted their result. That is data bias that a corporation wants to avoid. It has reduced their profitability.

Now, let’s assume the same process, with no internal corruption including careful well-intended procedures for collection, organization and input into their models. The data is the data as it was acquired, representative of the marketplace the company is trying to create product for. This process will likely identify the right product for that marketplace, provided the company’s models are good. These are procedures a company is likely to take. It is in their interest.

Now consider further, what if THAT data is biased? What if that product has elements of bias embedded in it but it is profitable and meets the corporate sales objectives. What is the company’s responsibility here? Have they done something wrong? Are they perpetuating or even deepening bias? Activists and purists answer “absolutely, Yes”. But you can certainly argue the other side, that the company is simply meeting the world, where it is, to maximize their profits. This is classic goal misalignment, the AI safety crowd knows that removing bias is good for society. The company doesn’t include societal goals in its evaluation process.

The AI safety crowd has two options. Argue that the removal of bias will result in MORE profit or argue that that company should sacrifice profit (or potentially not lose profit) by limiting the spread of bias. The former is difficult to prove. The AI safety crowd should be looking for this argument, because it is the ultimate winning argument (the proverbial win-win). However, I am skeptical that we can find it. It doesn’t make intuitive sense that a data set describing the marketplace, once altered to remove the bias of that marketplace somehow will be BETTER tailored to sell.

So that leaves us pleading for the merits of social good, which are likely at the expense of profit. That concept has significant meaning in the development of AI and ML on the corporate level.

  1. There is little or no incentive for a company to identify embedded, societal bias in advance,
  2. The identification of bias would require the devotion of resources, diverted from the more obvious mission
  3. The company would be responsible for adjusting/accounting for the bias and that requires a moral judgement that corporations are rarely equipped to handle
  4. The search for bias is neither easy nor obvious. A proactive company, willing to sacrifice profit, will even find this a daunting challenge.
  5. Bias can remain hidden for long periods of time

Given those challenges, I suspect that while companies may pay lip service to the challenges of bias, they are more likely to avoid the issue and focus on their substantial business challenges.

Only a regime of truly independent audit, with pre-established rules to identify and remedy, can succeed in removing bias from a company/dataset. There are simply too many hurdles with poorly defined benefits for the company to readily pursue a course of sufficient protection against bias. That is not to say that there won’t be times, markets or even whole companies that will identify benefits from the removal of bias, because there may be. But across the board, using limited resources and internal expertise, it is highly unlikely that the majority of corporations will prioritize bias.

Remember, this is not justification or endorsement of bias, but rather a wake up call for the AI Safety movement. This isn’t going to be easy. We must be pro-active to remedy bias. We must be creative. It is our responsibility to change the will of the marketplace to root out bias from our society (data) as well as from our corporations. Only when company’s profits are reduced for dealing in biased products or better yet, when company’s profits are increased for actively pursuing the removal of bias, will we actually see substantial gains against bias in all its forms.

Tackling Bias in Machine Learning, AI and Humanity

 

I nearly tweeted about a dozen responses back to Mr. Hamner, repeatedly however, I pulled up short. The nuance in this tweet is an enormous comment on where we are at in our discussion on AI and its interactions with humanity. Let me start with some thoughts and then we can dissect each one

  1. Direct reference to fixing ML bias being easier to fix than bias in humans
  2. Implication that overcoming bias in ML is more valuable than in humans
  3. Implication that some people are far more comfortable with machines than people
  4. A subtle undertone picked up by me, about corporate perspective, should bias be fixed (to be clear, I do not believe, this is not Mr. Hamner’s intent)?

Direct reference to fixing ML bias being easier to fix than bias in humans

I cannot imagine what the evidence for this argument looks like. I recognize that as a CTO, Mr. Hammer’s comfort level with machines is pretty high, but likewise an ethics professor is probably more comfortable working on bias at the human level. So are we talking about skill sets? Or is there a belief that rooting out Machine Learning (ML) bias is genuinely easy. If it is so easy, why isn’t it being done already and comprehensively. I reckon that it is not easy to identify and often hidden.

If it is easy to identify, then we should talk Mr. Hamner, because I would like to benefit from your expertise and partner with you in order to bring those benefits to the rest of the Machine Learning community on behalf of humanity. Especially to those groups who experience bias. It’s a worthy endeavor to be sure and I certainly hope that Mr. Hamner is right. If ML bias is easily identifiable then AI can go a long way to eliminating bias in our evaluation of data/markets and our decision making process. Humanity will have a lot to gain by eliminating bias.

Should you be wrong, and it is difficult to root out bias in our algorithms and in our data sets, then we are at the same place where humanity sits now, with institutionalized bias, but we are about to expand its reach. Not only would these biases be pervasive in our culture, but they would be codified in our artificial intelligence. I do hope that Mr. Hamner and others are well equipped to tackle this problem. ForHumanity stands ready to work with those who feel they have a good handle on this issue and to develop ways to make it a fundamental part of all AI and ML development.

Implication that overcoming bias in ML is more valuable than in humans

This implication made me uncomfortable. Not because I think Mr. Hamner is wrong, but rather I am concerned that he is right. One of the great things about ML and AI is that it often can be broken down into discrete building blocks. Fully observable data, transparent algorithms and dedicated processes may allow us to quantify the source of bias. If and when that is true, we may find it a fairly straightforward process to identify and readjust bias in our ML processes. However, today, we know that many deep learning processes are quite opaque to their designers. These technique have become so “deep” that their designers frequently are unsure why/how they work. This fact, for me, is worrisome, especially when considering bias. In these types of processes, if bias is introduced it may prove exceedingly difficult to remove. So, given the complexity of some of the ML going on today, we can be certain that perfect compliance is literally impossible. I remain optimistic, with Mr. Hamner, that in some ML we can identify and remove bias. Where we can, we should and it should be done post haste.

So then the question is, is it easier than overcoming bias in humans. Humans often obfuscate. Their data sets are not transparent, their algorithms completely opaque and their processes far from dedicated. But instead, humans have a will. They may even have a desire to change and seek out the elimination of their bias, especially when confronted with them. This is a societal decision, do we work with each other to face our bias, and then work to change them. This can only happen when we wake up and realize that all of humanity has EQUAL value. Minorities and Majorities, Each race, Each gender, Each sexual orientation, Each Faith or non-Faith, Each Political Party, Each Age and the list can go on and on. But we do not believe this today. Take our political discourse currently, each side thinks the other is either lunatics or ignorant. The answer is that neither is right.

The more value each member of our society places in each other member, the easier it is to eliminate bias. In fact, you would have changed the will of the people. Instead of hiding behind their bias, or worse yet, not even recognizing it, people will actively improve. Seeking out ways to eliminate their bias and increase the value that they can receive through equality. All very dreamy, I know, but it is a good dream and should be a goal.

This is all quite a long answer to the question of is ML bias easier to root out than human bias. But the answer is unsurprisingly, it depends. For certain situations and people, when confronted well by their peers and approached from an aspect of healing versus judgement, then I believe humans are easier to heal from bias, than any machine. Faced with the most difficult curmudgeon, who simply will not realize that all people have value, then the ML bias removal will be significantly easier and Mr. Hamner will be correct.

Implication that some people are far more comfortable with machines than people

This undertone to Mr. Hamner’s tweet makes me sad. Number one, I know this is a very accurate implication and two, I think it is increasing. If people are increasingly more comfortable with their machines, I believe they will actually be damaging their humanity, especially when that machine becomes the center of their focus. Even if our machines and our technology make our lives easier, do they make them better? From a microeconomic level, the answer is almost always “yes”, they do make them better. Otherwise, how else did that technology come into being? If you look at a cybernetically linked prosthetic that returns the ability of a person to have a hand and use it with their mind, with the same dexterity and functionality as before they lost their limb, there is beauty in that. A deep beauty that stirs the soul and endears the development of technology to the masses.

But from another more macro-economic perspective, our technology may have consequences we don’t realize. More importantly, I am certain that people, broadly speaking do not realize that much of our lives today are based on a MASSIVE assumption. An assumption that society will continue in its present form. Take GPS… a marvelous convenience to be sure. All maps at the ready, voice activated directions to allow you to keep your eyes on the road and hands on the wheel. Understandably, everyone uses it. But do people even realize anymore that knowing where you are, how to get from here to there and maybe most importantly, being able to read a map were once life and death skills. Failure to have those skills, meant certain death for some. It is easy to assume that our systems, our technology and our society will continue on in one direction never having a hiccup, a breakdown or worse yet a reversal. There are certainly scenarios I can imagine where many of the primitive skills, which were once common place to all people/society, may become required again. Putting all of your faith in technology might be easy and commonplace, but it doesn’t guarantee that it will always be there for you.

A society that eschews its human relationships and breaks down its sense of community is a fragile one indeed. Our technology is creating an illusion of self-sufficiency that is wrong. Each of us is completely dependent on a myriad of links in the chain that allows our existence to thrive. But like a chain link, it could be rendered useless, if even a single link were to break. And when your chain link breaks, who will be there to help? Not your technology. Will it be your community? Will it be your friends and peers? Will it be your neighbors?

I won’t dwell on the value of human relationships, as other have spent great time and effort documenting this, including a recent article by Brad Stulberg in NY Magazine. I recommend you give it a read.

http://nymag.com/scienceofus/2017/01/online-friendships-cant-replace-in-person-connection.html

I simply suggest that cultivating a robust community of human relationships takes considerable effort and it is an investment that will pay dividends in your-well being. Furthermore, it is likely to create a robust cushion for you if we do see disruption in society in its current form.

A subtle undertone, from a corporate perspective, should bias be fixed?

To be clear and fair to Mr. Hamner, I do not believe he was commenting on this idea. But as I began to consider ML bias, I realized that I doubt it is always in a corporate entity’s best interest to remedy bias in their AI. Now, before the pitchforks come out, let me be clear from ForHumanity’s perspective, YES, all bias should be removed. However, corporations may not be fully aligned with the best interest of society on this one. Let me explain.

If bias is removed from a company’s algorithms, the resulting decisions may not provide a product or solution that your customers actually want. Ostensibly, the data that was used in a company’s algos was identified as the right set to solve a problem for your customers. If that data set is altered to remove bias, the result might not be palatable to the customer, especially if they hold that bias. If their bias prevents them from employing the solution or purchasing the product, then the removal of the bias has hurt business. So while it is in society’s best interest to remove bias and to value all members equally, that is not how corporations act. Corporations, at least in the United States, have a responsibility to shareholders, not society. There are many examples in history, where companies have put their bottom-line ahead of the best interest of their community. We’d be foolish to think that will magically change now. Society and Corporations may be misaligned on the value to eliminating bias. So we certainly cannot rely on companies alone, to lead the way on the removal of bias from their AIs. We will have to make them do it.

Bias is wrong. In all its forms, in all its manifestations. Wherever it is found, it should be rooted out and changed. But this is society’s challenge. This is ForHumanity’s challenge. And it applies to our technology as well as it applies to all people. I want to thank Mr. Hamner for an extremely thought provoking tweet, whether he meant it as such or not. I hope that my thoughts are useful to all and that where appropriate, you will join with me to combat bias and to tackle the changes that AI & Automation pose for our humanity.