By Rishal Hurbans, Solutions Architect at Entelect | Founder at Prolific Idea
I hit a bird with my car the other day. I felt terrible. I don’t know why I felt that way. Some people might not be at all affected by the death of a bird. Why is that? What makes us respond the way we do in certain situations? What makes us think and act the way we do?
The short answer is: No one knows for sure.
No one understands exactly how the human brain works or what quantifies intelligence. Humans think they’re intelligent because they’re the dominant species. But that’s because we can’t yet comprehend what is more intelligent than us.
Yet we’re still trying to replicate the human brain through artificial intelligence (AI). But until we understand how our brains work, we are unlikely to achieve artificial “super” intelligence – and we may never understand or be able to comprehend what a greater intelligence is.
Artificial intelligence (AI), machine learning and deep learning are the new buzzwords on everyone’s lips.
But the concepts and methodologies behind these technologies are not new. Thanks to advancements in technology, we can use data, advanced algorithms and cheaper, more powerful computing hardware to achieve new and extraordinary things. Things like automatically describing images for the blind and detecting fraud in banking.
These are examples of narrow intelligence, the first of three categories of AI, and about as advanced as the technology is. For now.
Narrow intelligence refers to specific applications of AI. These algorithms perform a single task and can’t yet “learn” from past decisions about other topics to influence current ones.
Once they can learn and string different concepts together, however, we would have reached the stage of general intelligence, which is essentially what we understand intelligence to be today, in the context of the human brain. General AI can string learnings together and make decisions based on historical and real-time data. In other words, it can make a decision about one topic using experiences from another, related topic.
But super intelligence is beyond us. It’s smarter than humans, will possibly supersede our intelligence in seconds and is, therefore, something we cannot yet comprehend – and something many people – including Stephen Hawking and Elon Musk – are wary of.
If we develop something that’s smarter than humans, would it perceive humans to be bad? Could it be used for nefarious means? It’s difficult to say. But until we have a defined use case for super AI and a deeper understanding of intelligence, it will be a while before we achieve it.
Progression for a purpose
We do, however, have use cases for narrow intelligence. Lots of them. One reason why narrow AI is becoming more prominent within businesses is that computing power and data have made it feasible to experiment with AI – with the goal of making money and uncovering business opportunity.
A good example of narrow AI is the use of chatbots by banks and retailers to better service customers. Chatbots can quickly and efficiently respond to specific customer queries without any influence from a human.
But chatbots are flawed. They can only respond to programmed questions and they don’t yet understand sarcasm and depth in sentiment. They also can’t draw on past knowledge to interpret what someone is saying, to give them a coherent response. But, with advancements in natural language processing, we’ll get there soon.
Intelligence for a purpose
Organisations should avoid the temptation to implement AI just because everyone else is doing it because this puts them at risk of wasted time and investment.
But those that believe they have a use case for AI can take a bottom-up or a top-down approach.
With a bottom-up approach, businesses should identify a problem and then decide if AI can help solve that problem. AI generally works well for optimisation and automation. If there is a digital process that can be optimised, for example, then AI would work well. The key is that AI should be fit for purpose and must add value to the business, freeing up staff time and helping the business meet its goals.
But what if a business doesn’t have a use case for AI but still wants to experiment with it? In this instance, they can take a top-down approach, which adopts the concept of design thinking. In this approach, businesses create a goal and do everything they can to achieve it. If they want to use AI in that journey, they need to understand what data they have, consolidate it, analyse it and find trends that will inform their action plan.
Concerns have been raised about the potential of AI to make humans redundant in some jobs, or for it to land up in the hands of those with bad intentions.
These are valid concerns and, while it likely will replace jobs, AI will also spawn entirely new industries that will require skills that don’t even exist yet – much like what the internet did for the SEO industry, or what the industrial revolution did for machine artisans. And since we don’t yet fully understand the capabilities of AI, we don’t understand the potential either. It’s unexplored territory.
AI will create more responsibility for those in leadership positions because they will need to ensure that the intentions are always positive towards business and benevolent towards humanity.
It’s normal to fear the unknown. Every notable invention in history had the potential for abuse. But that shouldn’t stop businesses from experimenting. It’s the only way we’ll grow skills and knowledge in the industry and come up with new ways to solve problems and to innovate.
As the need for new applications of AI grows, we’ll need to start addressing the ethical questions but, for now, it’s helping us to be better, more innovative and possibly add business value. And I don’t see any harm in that.