Artificial General Intelligence and the Business World
While the term artificial intelligence has no fixed or inherent meaning and can instead be applied to a number of different topics, ideas, products, and services, a central theme throughout the history of AI is the idea of creating intelligent machines that have the ability to identify, combat, and ultimately solve a wide range of different problems. As seen in popular Hollywood movies during the past few decades, artificial intelligence has been portrayed in movies such as the Matrix and the Terminator as machines that have the ability to fight, reason, and communicate on a level that is not only comparable to that of human beings but also more advanced in some circumstances.
However, the scientific reality of artificial intelligence is that all attempts to develop general artificial intelligence (AGI) as of 2022 have all but failed. Defined as “the representation of generalized human cognitive abilities in software”, many individual researchers and software developers, in addition to multinational technology corporations such as Google and Microsoft, among others, have all made attempts to develop machines that can mimic human intelligence throughout human history. Nevertheless, all of these attempts ultimately run into the same inherent issues, as the ability of the human brain to handle multiple complex tasks simultaneously is something that scientists are still struggling to understand.
Narrow AI
To this last point, the overwhelming majority of AI systems that are currently being used within the scientific community and business world alike fall under the umbrella of narrow AI. For context, Narrow AI is defined as a “specific type of artificial intelligence in which a learning algorithm is designed to perform a single task, and any knowledge gained from performing that task will not automatically be applied to other tasks.” Put in simple terms, many AI systems are designed to accomplish a single task, and struggle to apply any knowledge or information that has been used to train these systems to any additional tasks.
To illustrate this point further, one of the most prominent examples of the implementation of AI and machine learning algorithms within our current business environment is the popular cloud-based typing assistant Grammarly. Grammarly functions to assist consumers in writing emails, academic papers, and training materials, among other things, as the application automatically reviews any text that a user writes when the program is active for spelling, clarity, punctuation, and delivery issues. However, Despite the numerous benefits of Grammarly as it relates to word processing and written communication, these benefits are limited to these pursuits, and cannot be applied to others.
Alternatively, “Ai-Da“, an AI robot that has been described as “the world’s first ultra-realistic robot artist”, has been trained to create works of art that are similar to those that have been created by actual artists. However, when Ai-Da was called to provide testimony in a mock trial that was conducted in the UK Parliament’s House of Lords this past week, the algorithms that had been used the train the AI struggled to answer questions in a consistent and coherent manner. This is due to the fact that Ai-Da was developed to generate artwork and not to mimic human language and communication.
The mysteries of the human brain
What’s more, the very level of understanding that human beings have regarding the functionality of the human brain has also impacted the development of AGI. For instance, while artificial neural networks are advertised as machine learning algorithms that function similarly to the neurons that are present in the human brain, this comparison is somewhat hollow, as virtually all forms of neural networks are merely approximations of the most basic functions of our brains, and even these approximations are still iterations of narrow artificial intelligence. Subsequently, a neural network that has been trained to predict weather patterns will not be able to simultaneously predict stock prices, as an entirely different algorithm will need to be trained to accomplish this different objective.
At the center of many issues regarding the advancement of AGI is the level of reasoning that human beings develop during their formative years. For example, children attend school beginning around 5 years of age and spend several years learning general skills that can be applied to multiple different life situations and pursuits, as a person who knows how to read and write in a particular language will be able to speak with other individuals, read newspaper articles, send emails, etc. This is in contrast to artificial intelligence as it currently exists today, as an AI machine would have to be created to mimic each of these human activities individually.
While artificial general intelligence may be developed at some point in the future, even the most advanced forms of AI around the world today still need some level of human input in order to function appropriately. Going back to the example of the AI robot Ai-DA, the machine struggled to respond to questions in an efficient manner even though the software developers that created the machine had received these questions ahead of time and trained it on prerecorded answers. In this way, progress toward machines that truly have their own thoughts, emotions, feelings, and intuitions seems to have progressed very little when considering real-world applications.
