Computers are amazing at some tasks and surprisingly dull at others. What gives? What makes a computer able to crush mind-boggling math problems but fail at even the simplest existential questions we have? If you’ve ever noticed that people have an easier time answering a math question coherently than a question about their feelings you might wonder how correlated human thinking is to a computer. Here’s a thought. What if we are having trouble making computers “smarter” because computers are already as smart as we are?
It’s an amazing puzzle to me that this question seems both completely false and strangely true. How are computers and people the same and how are we different when it comes to thinking? The first thing that comes to mind is the surprisingly nondescript “computers think in ones and zeroes” catchphrase. How do those ones and zeros, or binary, translate to anything? If we steer ‘round the computer science ice burg we might get to see binary in plain English.
Binary and Non-Binary Questions
A binary question is a question that restricts answers to just two categories. Responses requiring answers like right, wrong, yes, no, true, false, this, that and many others are for binary questions. For those of you that see the gaping hole in this comparison, don’t worry you are perfectly sane.
In a perfect world, if you ask a binary question it should have a binary answer. If this criterion isn’t met your question simply won’t make sense. That’s why, for example, when you type “do I have malaria” into your Google search you don’t get an answer that makes sense.
Our gateway to the confusion is when a non-binary idea is squeezed into a binary form. Here are some more examples:
- Is love unconditional?
- Does everyone have a purpose?
- Is wisdom better than intelligence?
- Is socialism evil?
- Does anyone on Capitol Hill really know what is going on?
All these questions are broken questions. They are non-binary questions packaged into a binary format. People answer questions like these because they sometimes have the contextual information to do so. Or, they know how to get it. However, the majority of the time people answer questions like this because we feel that the value of any answer is better than no answer. A quirk we passed down to computers. That is why your search returns millions of answers instead of, “That’s a dumb question, please try again”.
Most of us would attempt to answer a broken question anyway because we expect a back and forth exchange that may be helpful. If you knew ahead of time that your answer was limited to a single breath and no further exchange was possible, I am sure that many of you would cringe at the idea of just giving a yes or no answer without any further explanation.
What people don’t realize is that computers answer non-binary questions with a pretty neat trick (shown above). See, computers are phenomenally good at being consistent. Not right, just consistent. If you ask a computer a question it will give you the same best answer WE came up with over and over again for all of time. All we need is one person to be right at something once for the rest of humanity to benefit. I think it is an amazing deal.
Because of their remarkable consistency, a computer can trade-in a “hard” question for a few billion smaller ones. There are limits of course. Some questions require so many smaller ones that they aren’t worth asking. Those are put onto the quantum computing back-burner. Another limitation is that your question has to make sense. For a computer, this means that the question has to be able to be answered in some amount of binary questions. Outside this narrow scope, a computer is just a beeping silicon rock.
Thinking more human
Being consistent and logical is not the human modus operandi (MO). Most of the questions we ask and expect from each other are binary in form even though they are non-binary.
Just because your question is grammatically correct does not mean that it makes any sense at all.
The proliferation of computing and its use should be evidence enough to demonstrate our combined lack of prowess for logic and yet how important it really is to us. I have found myself plenty confused by some questions that sounded simple. It’s even been embarrassing. If like me, you have agonized over something like this then I hope you find some redemption in knowing that in some of those cases, it was just a broken question and your confusion was probably warranted.
Computers can only answer questions of any complexity by trading it in for many smaller questions. You and I can’t do that same trick. Giving a non-binary question to another person in binary form is like asking them the ratio of red pixels to blue pixels in a printed black and white photograph. Here are some things to watch out for.
Tips for thinking more human:
- Avoid asking/answering a yes or no question for any question that can’t be answered with a yes or no. Ask for clarification.
- If you have figured out the “reason” for something. Stop and challenge yourself to come up with 2 or 3 more. If you are feeling dangerous, try to rank them by a metric that makes sense.
- Understand that there is a difference between what people say and what people mean. Computers struggle with this because people do.
So, are computers as smart as we are? Don’t stress too much because this is just another non-binary question squeezed into binary form. The only thing interesting about this question is why the answer isn’t abundantly clear.
One big problem with binary thinking is that it is over-applied to problems that are non-binary. If the problem is even remotely complicated then a binary solution is non-sensical and even distracting. When a label like good or bad is put on a solution people stop asking questions that would otherwise force them to learn more about the issue in order to break it down and solve it.
A binary mask can obfuscate even the simplest problems making them impossible to solve. Binary questions might be a cup of tea for a transistor but they can cause havoc for us when they aren’t noticed. Likewise, a human’s greatest tool is the ability to infer from context. Those inferences don’t mean anything though if we don’t have the context or situational knowledge to back it up and or prompt more questions.