AI will often not as effective as a human doing the same job. It will not always know more or be more accurate. And it will definitely not always be more fair or more reliable. But still It can be used whenever you have an advantage over humans. And two Harvard scientists, Bruce Schneier and Nathan Sanders, have published a study in which the four keys point out: speed, scale, scope and sophistication. Understanding these dimensions is key to understanding the replacement of AI by humans.
The first one is speed. All those tasks, such as data processing, image escalation or patterns detection, An AI can make them faster than humans. Our advantage is in the details, in, once the information is screening, identify the gold nuggets. ”
“IA models can perform this task with incredible speed -Explains the study -, a capacity with important industrial applications. IA -based software is used to improve satellite and remote sensing data, compress video files, optimize video games with more economical hardware and lower energy consumption, help robots to make the correct movements and model turbulence to build better internal combustion engines. ”Those will be the most vulnerable works.
The second dimension of AI advantage over humans is the scale. The AI will be used more and more in tasks that humans can perform well in one place at a time, but that the AI can perform in millions of places simultaneously. A known example is the segmentation and customization of ads. Marketing professionals can collect data and predict what kind of people will respond to certain ads. This capacity is important from the commercial point of view; Advertising represents a worldwide billionaire market.
“The AI models can do this for each product, television program, website and Internet user -points out the study -. This is how the modern industry of advertising technology works. The bids in real time set the price of the ads that appear next to the websites that you visit, and the advertisers They use AI models to decide when they want to pay that price: thousands of times per second. ”The key will be to appeal to ads with a content that AI cannot carry out, such as emotion.
The third segment is the scope. AI can be advantageous when it does more things than anyone, even when a human could be better in any of those tasks. The generative AI systems such as Chatgpt can file conversations on any subject, write an essay defending any position, create poetry in any style and language, write computer code in any programming language and much more. A human being can do several of these things, but not all. And less at the same time. The response to the scope or versatility of the IAS will be the specialization.
“It is likely that companies continue to depend on human specialists to write the best code and the best persuasive text, but More and more will settle for AI when they only need an acceptable version of both“Affirms the study.
Finally, the last section is occupied by sophistication. IA may consider more factors in their decisions than humans, which gives them a “superhuman performance” in specialized tasks. The chess computer systems of the 1990s, as Deep Blue, for example, They succeeded anticipating a dozen or more movements.
But modern AI systems use a radically different approach: deep learning systems, built from multiple neural networks, They take into account complex interactions, often thousands of millions, among numerous factors. The reality is that the sophistication of AI provides value to many scientific studies, and its use in all scientific fields has grown exponentially in recent years.
“Identify obstacles in speed, scale, scope and sophistication – concludes the study – provides a framework to understand where AI provides value and, at the same time, where the unique capacities of the human species give us a lasting advantage. ”