The history of Artificial Intelligence (AI) began in the 1940’s when the first theories on machine intelligence were developed. But it was in 1956 at the Dartmouth College Workshop that the field of AI research was created by the most important fathers of AI, like Allen Newell, Marvin Minsky, and John McCarthy (who coined the name Artificial Intelligence, and organized the workshop which saw also the participation of Arthur Samuel, Oliver Selfridge and Herbert Simon).
The 60’s 70’s and 80’s are considered “AI winters” because not much progress was made, regardless of the optimism and plentiful funding, mostly because the computers were not powerful enough to deal with the amount of data necessary to create new models of AI applicable to different aspects of our culture. It was in the 90’s and early 2000’s that AI technology started to find new applications in many fields due to the advancement of machine learning, more powerful computers, and bigger data sets. Then in 2012 with the emergence of deep learning, AI improved significantly in the areas of image and speech recognition.
In the last several years, especially with the advent of ChatGPT (generative pre-trained transformer) developed by Open AI, and similar algorithms by Microsoft (Bing), Google (Gemini), Amazon (Amazon Lex, Polly, Q), and Facebook (Meta AI), it is now possible, with prompts or by voice, to ask them to assist us in many different ways.
Some of the earliest AI applications were game playing, first among them checkers in the 50’s. Soon after it was used by logic theorists, like Allen Newell, to mimic humans logical reasoning, and for language translation. Later, from the 1990’s to 2012, due to advancements in technology, like the internet, more powerful computers, and the advancement in deep learning, it became more widely used. In the last few years AI models have been utilized, for example, to write poems, songs, to create images, digital paintings, as financial advisors, for face recognition, to make autonomous cars, trucks, commercial and military ships, military and civilian drones, as well as smart robots able performing many different tasks in homes and factories. And let’s not forget about the AI chatbots now available to assist us in many different ways, from sales to entertainment, to banking. Just as useful, if not more, AI algorithms have also been employed in the arts to restore lost artworks, in languages to preserve those in danger of been lost forever, and when it comes to our cultural heritage, they made it possible to save millions of documents by digitizing and making them available to researches for many years to come. Nowadays we are very well aware of the profound impact of AI in health care, like its capability to analyze images to detect cancer early; in education for helping students research, write and edit assignments, for providing them personalized learning experiences, acting as a tutor able to adopt to each student pace and needs; for the climate, for monitoring and predicting changes by analyzing satellite images and thus help in the conservation and preservation of our flora and fauna.
If the above are useful applications of AI, the following are not. This should not surprise us. Like any other technology since the beginning of time, like fire or rocks, AI can be used also in unfair, illegal and unethical ways. Concerns range from privacy to discrimination, because biased data can lead to unfair treatment in hiring, in obtaining a mortgage, and in how the laws are enforced. It can bring about job displacement, as well as loss of jobs opportunities do to automation. It is a security risk when used for cyber attacks or autonomous weapons. AI is already used to create deep-fakes and to disseminate disinformation, which makes it very difficult to distinguish fact from fiction.
We should welcome all AI contributions that make or lives better, but we must also address its negative aspects, including its potential for the creation of biological weapons by rogue nations, or even by criminal and terrorist groups. In many places this problem is being addressed (as of 2023, 34 countries have some kind of AI regulation). In June 2024, the EU published “The AI Act,” one of the first and most comprehensive and strict policies on the uses and limits of AI in the areas of privacy, discrimination, cybersecurity, and automation. The US has not yet created a law addressing AI problems, essentially leaving it to the major companies that create or use AI to monitor themselves, which is not really what is needed.
The main worry with the US and allied governments creating laws requiring companies to adopt a much stricter approach to the development and applications of AI, making sure it is safe and fair before releasing it the the public, is that it would slow them down and fall behind in the AI race with competitors or enemies, like China and Russia, who put no constrains on their companies. Losing this race is just not an option for the major powers which see it as an existential threat. For this reason no one has not yet seriously proposed a global treaty, like those for chemical and nuclear weapons. Nevertheless, it is necessary both nationally and globally to start creating the basis for reaching a consensus on how best to control the most controversial, unfair, and potentially devastating AI applications.
The UN has undertaken several useful initiatives, but we know how effective it is when it comes to making real enforceable policies, especially now since the five permanent members of the Security Council do not agree on anything that really matters. Nationally, a good starting point for the United States government would be the creation of ethical guidelines for a fair, responsible, and accountable use of AI: to implement regulations to prevent its misuse, like surveillance, biased decisions, deep-fakes and disinformation which are so pervasive already; it can invest in education to prepare the next generations with new AI related jobs, or for loss of jobs. As some have suggested, it can propose a short work week, and create a universal payment system using some of the profits generated by a more efficient, autonomous, and productive economy.
Although these are all important problems that require solutions, the one I worry about the most concerns the future of “natural” humans, humans as we know them today—some will become transhuman, superhuman, enhanced by a combination of AI, genetic engineering, mind uploading, neuroprosthetics, and brain-computer interface (BCi). Will the next AI, AGI (artificial general intelligence) — which, with much greater, all encompassing cognitive capabilities than ours will, theoretically and practically, be able to control all levers of power and production, and thus all aspects of our lives, (and not just ours) — sooner or later decide it does not need, or wish, to worry for, or take care of ‘natural’ humans and their needs? Although alarmist, this is no longer an improbable scenario, not longer just dystopian science fiction, because in many ways fiction today is reality itself.
