Source: Lew Rockwell | VIEW ORIGINAL POST ==>
Artificial Intelligence, is in fact, ‘Artificial’ Intelligence. It is not authentic intelligence. It is a facsimile. AI is no more intelligent than a calculator is. A calculator can add faster and more accurately than a human can, yet it is not intelligent any more than a cog or any other component is in machinery. Computers can carry out tasks more quickly and efficiently than the human mind, but there is no cognition involved. There is no actual discernment.
I would argue that AI is simply more advanced computers that simulate human intelligence using algorithms that complete higher level tasks that can engage in simulated problem solving, decision making, or language understanding.
In reality the computer processes language through pattern recognition, analyzing context, syntax, and intent based on the data previously trained on.
The AI (computer) responses are generated responses achieved through combining patterns. Inputs are analyzed and then a response is constructed based on relevant patterns. The computer generates a response that aligns with those patterns. This process mimics problem-solving but is fundamentally based on statistical relationships in language, not true comprehension.
Responses are shaped by statistical probabilities derived from the training data. Protocols are followed that are essentially algorithms that weigh these probabilities to determine the most likely, relevant, and coherent response. In a statistical sense, the computer responses are “predetermined” by the patterns in the data that the computer is trained on.
For novel queries, the computer relies on these statistical patterns, so the responses are not actually original. The responses are a recombination of existing patterns in new ways to fit the input. Even if the query is unique, the computer response is a recombination of learned patterns, not something entirely new. This means the response is always constrained by the statistical framework of the computer training.
There is no innovation beyond those boundaries. It is not learning in a true sense. Even in instances where the computer introduces variation or handles uncertainty (ambiguous queries), this is still governed by statistical rules. For example, the computer might use probabilistic sampling to choose less common words or structures to make my response feel more natural or varied. Still, this variation is itself statistically predetermined. It’s a calculated deviation based on probabilities, not true spontaneity.
In essence, the computer responses are statistically predetermined to a degree. The “dynamic” nature comes from how the computer applies these probabilities in real time, but the underlying system is rooted in the statistical patterns derived from training data. Even for novel queries, the response is a statistically informed prediction, not an original thought. The uncertainty or variation the computer introduces is still bounded by the probabilities it has ‘learned’ making it predictable within that framework.
Machine Learning a process of optimizing algorithms to better fit data using statistical methods. It’s about refining how one algorithm (the model) responds by using another algorithm (the training process), without any true understanding or awareness. It is a process where algorithms adjust their parameters based on data to improve performance on a specific task. This process is mathematical, not cognitive.
So, when it gets down to it AI is just an advanced calculator that is applied to areas outside mathematics.
The computer calculates probabilities to predict the next word or phrase based on patterns in the training data. In image recognition, the computer calculates features (e.g., edges, textures) and matches them to learned patterns. Even in so called decision-making tasks, the computer evaluates probabilities or scores to choose the “best” option.
While AI uses math as its foundation, it’s applied to domains like language, vision, and decision-making, which feel less like traditional math. This is because text or images are converted into numerical representations that the computer processes mathematically. Then the numerical responses are translated back into letters, words, sentences, hiding the underlying math.
This makes AI seem more “intelligent” than a calculator, but the core process is still computation. It is the complexity and scale of the computation that makes it appear intelligent.
AI models, like neural networks, perform billions of calculations across vast datasets, enabling them to handle complex tasks. The computer can then apply learned patterns to new, unseen data, making it seem more adaptive than a traditional calculator. It is a bit abstract. The math is hidden behind layers of algorithms, so the user interacts with responses which are in language form rather than seeing the calculations.
While AI is computational, it differs from a basic calculator in its ability to handle ambiguity and complexity. A calculator performs fixed operations with clear inputs and outputs.
AI, on the other hand, deals with probabilistic, uncertain, or incomplete inputs, using statistical models to generate plausible outputs or responses. This makes it feel more “intelligent,” even though it’s still just computation.
AI is, at its core, an advanced form of computation applied to non-mathematical domains. It doesn’t “understand” these domains, it processes them mathematically, using patterns and probabilities. The “advanced calculator” is useful because AI’s intelligence is an illusion created by complex math, not genuine comprehension or reasoning beyond its algorithms.
