By Callan Abrahams, Principal AI Consultant, iOCO

Callan Abrahams

Despite the fact that AI is in its infancy relatively speaking it is developing exponentially, and while we haven’t yet made a machine intelligent enough that it can handle any general cognitive task in any setting, we have developed technology that can solve problems and build intelligence.

AI can be divided into two types: Weak (narrow) AI, focused on one particular problem or task, and strong (general) AI that focuses on building intelligence that can handle any task or problem in any domain. However, these terms are deceptive – weak AI isn’t feeble, and strong AI isn’t all-encompassing.

Weak AI is still incredibly powerful

Most of today’s AI applications are considered weak AI. Virtual assistants such as Apple's Siri and Amazon's Alexa, language translation tools such as Google Translate, image recognition systems, and recommendation algorithms used by platforms such as Netflix and Spotify are all examples of weak AI.

Designed and trained for a specific task or a narrow set of tasks, narrow AI excels in those particular domains. As a result of the fact that it operates under far more constraints than even the most basic human intelligence to perfect a task and perform it exceptionally, weak AI has powered many technological breakthroughs in just the last few years.

Some authorities believe that weak AI is easily the most successful realisation of AI to date. Smart assistants can set reminders, search for online information and control the lights in people’s homes thanks to weak AI and its ability to collect the right type of data. Learning user preferences, narrow AI can improve interactions through prior experiences and feedback.

The promise of self-driving cars is based on weak AI. Using deep neural networks to detect objects, determine distance, identify traffic signals and so on, driverless cars are being built on narrow AI’s focus on personalised, efficient awareness. Weak AI has already proven that it can outperform human talents in specific fields and is having a substantial impact on industries such as banking, healthcare, and customer service.

General AI

Strong AI is meant to imitate human intelligence. The idea behind general AI is that it should be capable of performing any intellectual task that a human being can do, and can learn from experience, adapt to new situations, and apply knowledge to different domains. With the ability to understand, learn, and apply knowledge across a broad range of tasks similar to human intelligence, Strong AI aims to understand context, make decisions, and solve problems in a way that is indistinguishable from human reasoning.

With a goal to achieve comprehension, adaptability, and independent thinking equivalent to humans, general AI is the type of artificial intelligence that has resulted in fear mongering and predictions that AI is going to take over the world. Thankfully, we haven’t yet developed a strong AI that can be used in the real world.

The closest we have come is large language models (LLM) and image generation models. The best known LLM, GPT-4, feels like Strong AI because it can be difficult to distinguish between AI-generated and human text. Today’s AI is not intelligent as most people would define intelligence, but it has been used to create some extraordinary applications. For example, iOCO has used LLM foundation models to build a regulation and compliance assistant for customers in heavily regulated industries, saving them valuable time and helping them avoid human error or missing tiny details.

Reaching AI’s potential

Many would say that the ultimate goal is Artificial Superintelligence, an AI system that surpasses human intelligence in virtually every aspect. Artificial Superintelligence would have cognitive abilities that far exceed the most brilliant human minds across all domains, and would be capable of outperforming humans in scientific, artistic, and problem-solving endeavours.

The distinctions between the different types of AI emphasise the scope of artificial intelligence's potential. Weak AI has proved itself as a vital tool in a variety of applications, and tech visionaries are actively working towards Strong AI.

As we move further into a future where AI is integrated into every part of our daily lives, it is vital that we start creating guard rails to ensure that AI solutions are used for the benefit of humanity, not to our detriment. Regulations will therefore become increasingly important in making sure that AI is used for good. We may be years, or even decades, away from full awareness and an intellect comparable to human cognition, but AI is advancing in leaps and bounds, helping us create a world where we are augmenting human intelligence effectively enough to debate the dangers – and benefits – of the different types of AI available to us right now.