Orrich Lawson | Getty Images
“Artificial Intelligence” as we know it today is a misnomer at best. AI is by no means intelligent, but it is artificial. This remains one of the hottest topics in industry and is gaining renewed interest in academia. This is nothing new. The world has gone through peaks and troughs in AI over the past 50 years. But what makes the current spate of AI successes different is that modern computing hardware is now powerful enough to fully implement some of the wildest ideas that have been around for a long time.
In the 1950s, in the early days of what we now call artificial intelligence, there was a debate about what to name the field. Herbert Simon, a co-developer of both the Logical Theory Machine and the General Problem Solver, argued that the field should be given the more bizarre name of “complex information processing.” This certainly doesn't inspire awe like “artificial intelligence”, nor does it convey the idea that machines can think like humans.
However, “complex information processing” is a much better way to describe what artificial intelligence actually is. In other words, it involves analyzing complex datasets and attempting to draw inferences from their accumulation. Modern examples of AI include voice recognition (in the form of virtual assistants like Siri and Alexa) and systems that determine what's in your photos and recommend what you should buy or watch next. It is included. None of these examples can match human intelligence, but they show that with enough information processing, we can do amazing things.
It doesn't matter whether you call this field “complex information processing” or “artificial intelligence” (or the eerily Skynet-sounding “machine learning”). A huge amount of work and human ingenuity went into building an absolutely amazing application. As an example, let's look at GPT-3. GPT-3 is a natural language deep learning model that can generate text that is indistinguishable from human-written text (though it can also be shockingly wrong). It is powered by a neural network model that models human language using over 170 billion parameters.
Built on top of GPT-3 is a tool named Dall-E that generates images of whatever fancy things users request. The updated 2022 version of this tool, Dall-E 2, allows you to go even further by allowing you to “understand” even the most abstract styles and concepts. For example, if you ask Dall-E to visualize “an astronaut riding a horse in the style of Andy Warhol,” it will produce a number of images like this:
Dall-E 2 does not perform a Google search to find similar images. Create an image based on an internal model. This is a new image built solely from mathematics.
Not all applications of AI are this innovative. AI and machine learning are being used in almost every industry. Machine learning is quickly becoming a necessity in many industries, powering everything from recommendation engines in retail to pipeline safety in oil and gas to diagnostics and patient privacy in healthcare. Masu. Not every company has the resources to create a tool like Dall-E from scratch, so there's a lot of demand for affordable toolsets. The challenge of meeting that demand is similar to the early days of business computing, when computers and computer programs began to proliferate. of We need a technology business. While not everyone needs to develop the next programming language or operating system, many companies want to harness the power of these new research areas and need similar tools to help them do so. is.
