Yet Another Problematic Polling Fallacy

 

One of the conundrums that is confounding the economists, well actually everyone is the disconnect between the US markets/economy and the public’s opinions about the economy. But it is not just the economists who are puzzled; it’s the financial experts and industry, and the news media and journalists, the politicians, and especially the political strategists. Well, then there is the public but they’re often confused about the economy, which isn’t exactly unusual.

Now, add to this the politics of an election cycle and the confusion of the conundrum is complicated by the chaos of confounding one causal factor with another. Want to know what the two causal factors are? They are: political party and the economics of inflation. You may not think that political party, or politics for that matter, has anything to do with the state of the US economy; and you would be right. However, that doesn’t mean that politics doesn’t have anything to do with polling about the US Economy. In a weird way, economic theory is non-partisan and yet partisan politics is an influential factor in economic policies, economic conditions, and public perception about the economy.

What does this mean about public opinion about the US economy? It means that economists and economic commentators are highly likely to be misreading and thus misinterpreting what is driving public opinion and how it corrupts their forecasts and predictions. It means we are contaminating our perceptions and understanding of the US’s Economy with a factor that isn’t accounted for in any competent economy theory. Unfortunately, this is assuredly happening, and it is producing bad decisions by, well, everyone. I don’t find this unusual, in fact, I have come to expect it.

But how to provide information that might help a small percentage of the population, even some economists, to see the problem for what it is; that’s the quandary.

Let’s try the following.

Suppose we separated the public into three groups, along a political spectrum. Just for simplicity purposes each group will be basically equal in size. One group are designated as Republicans and as 33% of the population. Another group is designated as Democrats and is also 33% of the population. The final group is Non-aligned and is 34% of the population (their extra 1% is just non-partisan simplicity).

If we assume that 80% of the Republican and of the Democrat groups will voter exactly opposite to each other and remaining 20% will vote a middle choice response on an economic question because of political views than we can create a view of how that would impact understanding polling data that is influenced by such a political factor. The third group’s views on economic questions could be divided into equal parts/numbers for the poll’s results.

A pretty simple model of how partisan politics might impact polling numbers/results.

Now, we are ready to conduct a theoretical poll. Here’s the poll question: “How do you rate X’s handling of the US Economy?” Respondents are given a choice of: Good, So-so, or Bad.

What then are the “results” ?

·   37.6% of the public approves (a rating of Good) of X’s handling of the US Economy.  

·   24.8% of the public rate X as doing So-so. 

·   37.6% of the public disapproves of X’s handling of the US Economy.

If the So-so rating is not considered a favorable rating, then X is not doing well with a partisan influenced polling reaction. But what if the partisan factor were discounted? What would that mean? This is calculated by eliminating the either one the partisan sides; and when we do that we get:

·          56.1% approve.
·         27.1% are So-so.
·         16.7% disapprove.

Clearly the political views have a significant and substantive impact on the poll’s results. This is a problem for assessing and making economic policy decisions. Politics as it usually does, pollutes the process of intelligently using polling information to gauge how the economy is being perceived. And in a world where public perception contributes to economy reality, actions, and processes nothing good will come of it.

What is X’s real rating then? It’s likely to be found somewhere between the two extremes, but where will depend upon just how inanely partisan and informed each of the three population sub-groups is.

Can Unintelligent Politicians Match The Sub-Par Intelligence of Artificial Intelligence?

 

The long-awaited threat to human existence from Artificial Intelligence (AI) is here!

Finally, we have something that we can use to measure human intelligence against. We have been limited far to long by having to rely upon our own self-referential abilities as the yard-stick. We created a notion of the Intelligence Quotient (IQ) to measure some general metric of where a human falls within the population of humans. It’s not a hugely useful or reliable measure which is pretty consistent with human nature. The ‘Natural’ Intelligence Level (NIL) of humans has been an elusive quantity and quality since the first human asked their self the salient question about another human: “Is that person really that ‘stupid’?” *  Before you take immediate offense, I want to alert you that I will shortly bring up the issue of politicians, government, and corporate executives/leaders discussing how to ‘solve’ the problems/risks that AI presents in many contexts.

* Note: The term ‘stupid’ is likely to be considered offensive by some, so please forgive me for its use. I apologize for not being able to determine what other term would be as clearly understood and recognized by others and possessing the attribute of not being seen as just or more offensive. I thought about listing some of the other terms considered, except that just creates an even larger issue. Suffice it to say that I could find none that would be as easily able to fulfill the requirements of Information Theory.

Artificial Intelligence (AI) has apparently come close to passing the Turing Test initially proposed in 1950 by Alan Turing. For those unfamiliar with the Turning Test, it posits that if a person cannot tell if they are interacting with another person or with a machine (an AI in this context) then the AI is has passed the test, i.e., exhibits an intelligence equal to that of a human. How recently AI or a machine can be considered to have ‘passed’ the Turing Test is debatable both in terms of when, in what context, and compared to what human(s). There are situations from a decade or two ago that some claim the test was passed. However, defining human intelligence is not a simple thing to do, and even if I attempted to do so it would not be agreed to nor accepted by the vast majority of those who might be considered knowledgeable. It absolutely would not be accepted by most people.

Regardless of whether AI has or hasn’t passed the Turing Test, it is currently capable of fooling a significant portion of the public on any number of things. It is also capable of being used by humans for good or ill purposes. This is where the current risks and threats from AI come from for the foreseeable future.

This brings us to the question of the day. How likely is it that politicians, government entities, and AI corporate leaders are going to be able to provide the answers, solutions, and policies that will effectively eliminate (or more realistically mitigate) the risks, threats, and harms that AI technology will produce? This is the appropriate question because we have empirical data upon which to base an assessment.

To manage, regulate, control, or adapt how AI is used and how it impacts people there is another technology which has preceded it that give us a valuable insight into how well our politicians and technology companies & leaders have done in managing, regulating, controlling, and adapting it. In fact, we have several but let’s just focus on Social Media. I think we can all agree that the following assessments regarding issues around social media are reasonably good predictors of how well politicians, government, and technical entities will do with respect to AI.

All these entities have done very poorly when it comes to social media. The problems surrounding and embedded in social media are many and mostly unresolved. For one thing, the problems have been very much enabled by technology and the companies engaged in providing social media. Politicians and government entities are completely lost when it comes to understanding the problems and certainly have no ideas of how to address them.

When it comes to AI, this propensity for inaction and failure will be exponentially worse. This does not bode well for the public, nation, or the world being competently protected. This is not because the issues that must be addressed or the problems that have to be solved are exceptionally difficult. Many of the issues and problems are not hard at all to deal with and to benefit from. This is true partly because the same is true for the even simpler problems and issues related to social media.

The underlying problem isn’t AI or social media, it is that we don’t have individuals involved in finding the solutions who possess the skills, competencies, and perspectives to do the required problem-solving tasks. If you do not know how to understand a problem, you are relying upon just being lucky in what you choose to do or not do.