It is an interesting point in the history of human evolution and society. We have Black Box AI systems literally running our lives, and we do not know how they work. We have Black Box Artificial Intelligence computer programs running everything from Cancer analysis to Google’s Search Ranking Algorithm. Still, we are not sure how AI comes up with its answers.
So how dangerous is it to let a computer model (program) like Black Box AI that we do not understand run significant institutions in our lives or affect a Facebook or Google News feed? If you don’t believe something like this can negatively affect your life, look around. Is everything ok?
Today, much of society increasingly uses AI systems to make decisions but is often unaware of how the decision was made. So if you want to learn more about AI Black Box Systems, how it works, and how it affects you, please keep reading.
How Dangerous Is Black Box AI?
How Dangerous Is Black Box AI? Well, that would depend on who you ask. To some, it is not understood how neurons interact with one another. To others, it is an unacceptable situation.
Decisions were made long ago to run with systems controlling lives that we do not understand. But what are the ramifications of such decisions? Generative Adversarial Networks (GAN) AI generate images when given a random input of numbers. Sometimes the pictures are not good, but often they create photos of faces that don’t exist and are indistinguishable from the real thing.
These types of AI programs also produce news stories that are presented as if they were written by a real person or attributed to one.
During the 2020 U.S. Election, the news media was embroiled in many issues involving or instigated by social media, like the Cambridge Analytica–Facebook scandal. American media and social media news stories were found to be fake or put into place by foreign entities.
Because of the Black Box AI effect on issues like social media running with stories that were not true (Fake News), no one knew what was going on and who was instigating it. Eastern European troll farms and China affected over 140 million Americans per day at times.
In our opinion, Black Box AI is dangerous and can cause anything from mental anguish to the possibility of a civil war in the United States, all in the pursuit of corporate profit. Cyberwar is a clear and present danger, especially if they are using our own social media against us in this country.
Black Box AI and Social Media
The documentary movie investigation “The Social Dilemma” discusses how Artificial Intelligence secretly controls and manipulates our lives. It also talks about how Facebook, now Meta, has lost control of its algorithms at times because they do not know how they work and what they are doing.
The Social Dilemma documentary also discusses the profit motive and how it is an almost impenetrable wall created by the wealthiest companies in history. There are trillions of dollars involved in these scenarios, and the bottom line is advertising and the average person’s attention grab.
This Black Box AI deception is not a left-wing or right-wing political issue. It is an issue of “Chaos” caused by a technology we created. We are not sure how it works, then recklessly releasing it on society as a financially motivated experiment and just wait and see what happens.
Artificial Intelligence is a Tool
Artificial Intelligence is an incredible tool that, if used in the right way, not only improves our lives but can do amazing things in our future. It is used in agriculture, cybersecurity, education, and healthcare, to name just a few industries.
AI is also being used in industries and areas like workforce augmentation, autonomous self-driving vehicles, law, and the criminal justice system. Some of these are gray areas that concern the populous with losing their jobs because machines or robots will eventually replace the human workforce.
There are also many concerns with AI and privacy with law enforcement and data bias being fed to AI for its training. The reality is that all of these areas are true, and whenever there is Black Box AI involved, we just don’t really know or have access to the information within these systems to get to the truth.
What is Black Box AI?
Some forms of AI or Artificial Intelligence work with Neural Network Deep Learning Algorithms in computers or machines so complex they are called Black Box AI. Once trained, neural networks learn on their own. Although the people utilizing the Black Box AI know the information or “data” input and output answers, they do not understand how the computer came up with that answer.
The Black Box AI, Neural Network algorithms, can become so complex that no human could work through all the variables involved in the answers or making a prediction. There can be millions and billions of Neural Network data parameters or criteria combined to make decisions.
In the social media arena and AI in general, only a few individuals know what is going on inside a Black Box AI algorithm. Once these individuals leave a company, it can become chaotic for the corporation or entity to try and fix or adjust an AI program that they do not understand how it works.
Complex Artificial Intelligence Systems Explanations
Describing Artificial Intelligence Systems to the average person is complex. Suppose you are a sizeable multi-billion-dollar corporation utilizing Neural Networks to operate your business. In that case, the decision is sometimes made based on the cost assessment for profit against the future cost of a lawsuit. Black Box AI systems prevail in certain areas and situations because, frankly, it is just easier.
You don’t have to know how electricity works to turn on a light switch. Why should anyone have to understand how an AI algorithm works? If the management team weighs the pros and cons of a lengthy legal battle against what they can financially profit often, the company will take that gamble.
Black Box AI is any artificial intelligence system whose functional design and description operations are not understandable to the user or others. If a Black Box AI, Neural Network, has over 100 billion parameters, not even the data scientist and architect who created it can tell you what is going on inside and how it got its answers.
Veil of Secrecy
We can only scratch the surface of the veil of corporate secrecy involved with Artificial Intelligence. Yes, some are open-sourced and sunshine and rainbows, but even Google has deals with the NSA and CIA that we will never know.
In his book on Amazon, “The Black Box Society: The Secret Algorithms that Control Money and Information”, Frank Pasquale describes Black Box Artificial Intelligence’s effect on society.
“Secret algorithms can make (or ruin) reputations, decide the destiny of entrepreneurs, or even devastate an entire economy. Shrouded in secrecy and complexity, decisions at significant Silicon Valley and Wall Street firms were long assumed to be neutral and technical. But leaks, whistleblowers, and legal disputes have shed new light on automated judgment. Self-serving and reckless behavior is surprisingly common and easy to hide in code protected by legal and absolute secrecy. Even after billions of dollars of fines have been levied, underfunded regulators may have only scratched the surface of this troubling behavior.”The Black Box Society: The Secret Algorithms that Control Money and Information, Frank Pasquale
Industries or Companies Related to or Using Black Box AI
|Artificial Intelligence||Education||Media / Press|
|Autonymous Self-driving Vehicles||Electronics||Military|
|Cloud Computing||Healthcare||Social Media|
|Corporations||HR / Recruiting|
|Criminal Justice System||Industrial Applications||Tiktok|
Bring in the Lawyers
CSET, the Center for Security and Emerging Technology, based at Georgetown University’s School of Foreign Service, in the first of a series of policy briefs titled AI and the Future of Disinformation Campaigns ”Part 1: The RICHDATA Framework-a Disinformation Kill Chain” and ”Part 2: A Threat Model”.
These briefs examine how people can exploit advances in Artificial Intelligence to enhance and automate disinformation. They also describe how AI can improve current techniques to increase disinformation campaign’s in their speed, scale, and personalization. These directly relate to the necessity for transparency in AI and responsibility.
In an article by Andrew Burt from the Harvard Business Review called “The AI Transparency Paradox,” he discusses industry and academia calling for more transparency into artificial intelligence models.
Burt also discusses in his article how exposing information about machine-learning algorithms can make them more vulnerable to attacks and how it has been demonstrated that people can steal entire algorithms on explanations descriptions alone.
He comments that “it is common for legal departments to manage risk assessments and even incident-response activities after a breach. The same approach should apply to AI”.Andrew Burt- The Transperacy Paradox
What is a Neural Network?
If you are talking to your Smartphone, that’s pretty much a neural network. Image Recognition, Voice Recognition, Gmail intelligent sorting, and suggestions on Amazon are all Neural Networks.
Artificial Neural Networks (ANN), or just Neural Networks, are fundamental problem-solving rules or algorithms in Machine Learning. ANNs are loosely inspired by the human brain and are a network of connected units or nodes called Artificial Neurons and are organized in layers.
Neural networks are a series of algorithms that make predictions from historical training information data, such as say on or off, night or day, or red or green. With each prediction repetition, it discovers if it is right or wrong. Each time it is wrong, it adjusts to make a better prediction the next time. After hundreds, thousands, millions, or billions of repetitions, that answer gets more accurate each time.
Neural Network Simplified Component List
Neural Networks are complex, and there are many components, levels, and settings, especially in Black Box AI scenarios. This list is a simplified explanation of some of the Neural Network features.
An Artificial Neuron is a unit or node and is a mathematical function based on a biological neuron. Neurons can both store and process information on their own so that there is no need to retrieve data from memory for processing like in traditional computing. Neurons have one or more inputs and are separately weighted. Neuron inputs are calculated and passed through a nonlinear function to produce output.
A Perceptron is an algorithm or set of instructions that enables neurons to learn and process elements in a training set one at a time. There are two types of Perceptrons, a single layer, and multilayer. Perceptron is a linear classifier that produces an output organized in a binary value between 0 and 1. It makes its predictions by combining a set of weights with the feature information.
There are many kinds of Neural Networks, and each one has many different parameters set with its design. Parameters are the settings that define how AI performs a job. Parameters form the Neural Network information input and answer output layers and are the neural network components that make predictions. Some complex neural networks have over 100 billion parameters.
There are three types of layers in Neural Networks: an Input Layer, an Output Layer, and a Hidden Layer. A hidden layer is between an input and output layer, and these layers are connected to each other and can be as simple as two layers, and there is no maximum number of layers in a deep neural network. The four most common neural network layers are Fully connected, Convolution, Deconvolution, and Recurrent.
Why Are There Hidden Layers in Neural Networks?
A hidden layer is between an input and output layer, as described before. In Neural Networks, neurons in a hidden layer perform calculations using some (or all) of the neurons in the network’s last layer. The neurons go down through the layers, and in that way, they mimic human brain neurons. The hidden layers are where the Black Box Magic happens.
Neurons in the hidden layers of neural networks are activated for an input value like a picture of an ear that answers specific questions of a whole elephant description. Hidden layers sit between the input and output layers and allow the neural network to be broken down into specific data changes. Each hidden layer purpose is specialized to produce a defined output.
Hidden layers are common in Neural Networks and are where neurons take in a set of information and generate an answer to the output layer.
It seems that there has been an odd black hole of information about Black Box AI since 2020. The pandemic has caused widespread apathy across society and the workforce and leaves the question, “What is really going on?”
Black-box models will likely continue to be permitted when it is obviously unsafe to use them. As the acceleration of computer science moves into Quantum Computing, we still don’t understand how our basic Machine Learning Neural Networks are controlling our lives, let alone answering complex questions.
It will all be protected by corporations and lawyers and hashed out in the end after a catastrophic breakdown of something, and hopefully, it is not society because unsupervised AI Algorithms are fanning the flames of discord. Dissension between all humans because of simple data biases passed over for profit.