AI: Understanding What It Means & What the Future Holds

In part one of this three-part series on AI, I will define AI, look at its history, and analyze the factors that have caused the recent interest in all things AI.

The life we live these days is increasingly digital and, in the course of a normal day, we seamlessly use several innovative services, including Google Translate to overcome language barriers, Amazon’s Alexa to help with tasks such as setting alarms and listening to music, and Netflix for streaming TV shows and movies on demand, in many cases which are recommended to us.

When we use such services, what we don’t realize is that these and many other similar ones are examples of companies using AI to develop more user-friendly products and services.

Early implementation of AI

The concept of AI is not new. As far back as the 1950s, AI has been in use in some form or other. Figure 1 explains the various stages that AI development has gone through over the past several decades. It is however, only in the recent months, with the advent of tools just as ChatGPT (from Open AI) and Bard (from Google) that the term AI had caught on in the truest sense among the general public.

Figure 1: A brief history of the developmental changes of AI

Early work in the domain of AI in the 1950s involved the development of Neural Network AI, which essentially focused on techniques that helped machines mimic basic human intelligence by using simplistic if/then type rules. This branch of AI focused on doing one type of task, and trying to master that based on a limited amount of data inputs.

The next big developmental phase of AI came in the 1980s, with Machine Learning and the use of complex statistical tools by machines to improve at tasks based on their past experience. This was the advent of machines “self-learning” and improving, marking a big leap in AI development.

Future development trends of AI

Then came ChatGPT.

ChatGPT is an advanced language model trained on large text datasets. It is specifically designed to interact with users in a chat format and can answer (or hold a conversation) in a human-like format.

Launched in November 2022, the explosive growth of ChatGPT has been nothing short of phenomenal. Within just a year of launch, the AI-based chatbot had amassed more than 180 million users and over 1.5 billion visits per month.

What sets tools such as ChatGPT (and Bard) apart from the rest of AI tools is that they create or “generate” new data (such as answers for questions) based on the input data they have been trained on. Thus, they belong to the next phase of AI development, called Generative AI. Figure 2 highlights the various stages of AI development within the domain of Deep Learning.

Figure 2: Developmental stages of deep learning AI

Narrow AI is currently the most prevalent type of AI. It’s only good at certain tasks, but can perform these tasks often at levels that surpass a regular human (or team of humans) even after years of practice.

Then comes General AI or Strong AI, which is the closest AI comes to current human intelligence. Although still in its nascent stage, this type of AI is able to perform various tasks, sometimes simultaneously just as humans do. ChatGPT is an early example of this. This type of AI lends itself to improved personalization and customization, is very automation friendly, and can reduce costs. It manages to achieve this by continually learning, teaching, and improving itself.

The next phase in the evolution of AI, which has yet not been achieved, is “Super AI”, which in, theory will surpass human intelligence. However, what exactly it means and what this type of AI can achieve is yet to be fully explored or understood.

Factors leading to the current AI boom

Even though AI has been around for many decades, the last year or so has witnessed growth in the interest and development of AI like never before, making it impossible not to think, “Why now?”

There are several intricately overlapping factors that together are creating an environment that is optimal to support a flourishing AI ecosystem. Some of the key reasons are described in Figure 3.

Figure 3: Factors leading to the current AI boom

 Source: Pulse

There are four major trends fueling the current AI explosion:

Computation Power Doubling è The increase in computational power has enabled the development and deployment of AI technologies that was once thought to be impossible.

Global Data Doubling è The amount of data globally has been doubling every two years, with the total volume of data expected to reach 175 zettabytes (175 trillion GB) by 2025. This wealth of data helps AI models better understand context of any situation and ultimately support the AI model to undertake related actions or outcomes.

Rapid Advancements in Hardware è Improvements and lower costs in graphic processing units (GPUs) and tensor processing units (TPUs) have made it possible to train AI models more efficiently and effectively. An example of this is how system-wide optimizations have helped reduce the cost for GPT 3.5 Turbo model by 90% over GPT-3.5.

AI Training Costs Decreasing è AI training costs have reduced drastically, making it more accessible and affordable for businesses and researchers alike to use and train new AI models. As AI training becomes more affordable, experiments with larger models and more complex algorithms can be performed, thus expanding the horizons of what AI can achieve.

Conclusion

The world as we know it is changing rapidly and, in some ways, quite drastically. The traditional boundaries of human achievement are undergoing a seismic shift. AI is enabling machines to do tasks that until recently only humans – and highly trained ones at that – could do. This means that some skills that were once very valuable will be no longer useful. On the other hand, AI will open up new horizons of growth and opportunities that will require today’s employees to upskill to remain relevant. And this is an ever-evolving domain, with new developments constantly reshaping the future landscape.

Organizations that understand the significance of this change — and act on it first — will be at a considerable advantage.

Disclaimer: Any views and/or opinions expressed in this post by individual authors or contributors are their personal views and/or opinions and do not necessarily reflect the views and/or opinions of Huawei Technologies.