RAVI VENKATESAN takes us on a journey through the world of Artificial Intelligence, charting its history and developments, and also exploring where we are headed.
Part of my job as the CEO of a technology company is to keep track of trends and technologies that could be incredibly transformational. This knowledge is critical to formulate strategies for our business. To use a hockey analogy from Wayne Gretzky, it is to “stake to where the puck is going.”
Artificial Intelligence (AI) is definitely a major trend today and has probably had more historical cycles of being sold as transformational and not living up to its promise than any technology in the last century. It all started with early neural networks in the 1940s and 50s that looked to make a computer mimic a human brain and its mental functions. In other words, the endeavor was to construct an electronic brain. The benchmark for success was established by Alan Turing in a 1950 landmark paper with what is now popularly known as the Turing test. In simple terms, if you interact (perhaps chat) with a computer, but are unable to tell the difference between this experience and chatting with a real human being, then the electronic brain has passed the Turing test.
So, what is AI? The fundamental building block of AI, or an “electronic brain,” is a neural network that mimics the kinds of neural networks our brains have, and in essence has a “learning mechanism.” I remember writing neural networks in languages like Prolog and C as a student. The fundamental difference between this type of computer program and other purely computational programs or software, was the aspect of being able to learn, potentially predict, and come to conclusions.
There was a lot of hype at that time about how AI would change the world and dominate fields as diverse as predicting the weather, to picking winning stocks, to automating architecture and design for buildings. However, very little of this came to pass. Funding dried up when people realized that too much expensive computing power was needed and the economics just did not work out.
The next generation of AI came in the late nineties when IBM developed Deep Blue, which beat the reigning world champion Garry Kasparov at chess. In 2005, a Stanford robot drove autonomously for 131 miles in the desert. These developments revived interest and funding again. So far, AI still needed very large-scale computing power and bespoke computing equipment – translation, a lot of money. This generation of AI can be called purpose-built AI. It was still in the domain of large companies or governments.
The next big move was Google’s DeepMind, which beat the reigning world champion at the GO game, much more complex than chess, and used “deep learning” techniques such as deep neural networks. This generation of AI can be called “deep learning general AI.”
The current generation of AI, with ChatGPT as a prominent example, is way more general purpose or multi-purpose and is called generative AI. It has the quality of not just arriving upon a result but also explaining the derivation of the result. It is thus far the closest to human intelligence that AI has reached. Carrying out a conversation with ChatGPT feels like talking to a very intelligent and articulate human being. With computing costs continuously decreasing, and “big data sets” to train the AI becoming more easily available, the improvement in AI with ChatGPT has been accelerated.
To summarize these generations, look at the graph below that plots Purpose and Cost:
![](https://cdn-prod-static.heartfulness.org/chatgpt_gamechanger3updated_4d03a5ff84.jpg)
So, is the promise of AI finally being delivered? It is safe to say that AI has finally reached a point where many applications have become economically feasible and hence will scale. Autonomous driving, facial recognition, improved customer service by scanning knowledge bases and using chatbots, authoring marketing content, writing code for software programs, etc., are some of the applications that ChatGPT is already doing successfully. It is the only technology that has been adopted by 100 million users in just two months after launch.
![](https://cdn-prod-static.heartfulness.org/chatgpt_gamechanger4_d4468f9f6b.jpg)
However, ChatGPT still does not conclusively pass the Turing test, so there is still a long way to go. A bigger question, often raised, is if AI is conscious. To answer this, let’s look at the various mental functions of a human being:
- Intellect – this is where logic and reason live. Prior to ChatGPT, AI had already developed logic, now reason has been incorporated as well.
- Ego – this is the sense of self, and while it can be programmed artificially, the way it develops in a human being has not been mimicked in AI.
- Mind – this is broader than just the brain, it includes the ability to process inputs from multiples senses (sight, sound, touch, taste, and smell) and also from internal sources (contemplation and thoughts). It includes the ability to imagine and create. AI is starting to mimic some of these functions. ChatGPT is the first AI that also “hallucinates” and can undertake creative processes such as authoring a poem, creating music, etc. However, the evolution of this aspect is at a very rudimentary stage.
- Consciousness – what forms the base for everything else in a human being’s mental functions can be described as consciousness. There are various levels of consciousness based on the sophistication of the being – humans, plants, or even stones, all have consciousness but of different levels. In that sense AI does have consciousness, but it is of a different kind than human beings. It is based on the programming and inputs provided to the AI and can of course evolve over time.
In summary, there have been multiple generations of AI going from very specific to very generic, and very expensive to very inexpensive. ChatGPT has been a giant leap from prior generations, and in some ways is democratizing the access to this technology by making it available to millions of regular people. AI has still not passed the Turing test and is not yet close to replicating all of our human mental functions, however the technology will continue to evolve, creating both opportunities and threats for the human experience.
Illustrations by ANANYA PATEL
![Ravi Venkatesan](https://cdn-prod-static.heartfulness.org/Ravi_Venkatesan_f84e97198e.jpg?&q=10)
Ravi Venkatesan
Ravi lives in Atlanta, Georgia, and is currently the Chief Executive Officer at Cantaloupe. He is also a regular public speaker and public speaking coach. He has been a Heartfulness meditator for over 20 years and is passionate about app... Read More