5 AI myths and why you shouldn’t believe them
There is more mystique surrounding AI than any other technology. Here, we debunk 5 of the most common AI myths you may believe.
Artificial Intelligence is arguably as significant as the discovery of electricity or the invention of semiconductors. AI is still relatively immature, but we are already seeing how it impacts all our lives. Virtual assistants are commonplace, even being embedded into other objects, like headphones, TVs, and cars. Facial recognition is used in more and more locations to identify known trouble-makers. Medicine and healthcare are being transformed by the use of diagnostic AI models. And, of course, AI is revolutionizing business and industry. The problem is, as AI gets more prevalent, a plethora of AI myths have grown up. I want to debunk 5 of the more pervasive AI myths.
AI will lead to everyone losing their job
This is probably the single most prevalent AI myth out there. Understandably, people are afraid that AI will lead to job losses. After all, one of the biggest selling points of AI is that it transforms business efficiency. As a result, you often see headlines like “The impact of artificial intelligence – widespread job losses” (from IT for All, July 1, 2020).
However, the reality is far more subtle. AI usually relies on machine learning to provide its “intelligence”. Most current AI systems can be described as narrow intelligence. That is, they do one task really well, but are not able to adapt to doing other tasks. Deep learning offers the possibility of general artificial intelligence, but it requires enormous computing power.
The upshot is, AI can only really replace certain roles, although it tends to do these extremely well. For instance, truck drivers have an extremely high risk of obsolescence, thanks to advances in autonomous vehicles. By contrast, AIs will never replace teachers, nurses, or other caring professions. There is a related AI myth that only low-skilled jobs are at risk. However, that is also false. In fact, it can be argued that artificial intelligence is effectively autistic. So, the jobs at the lowest risk are those that require emotional intelligence.
A superintelligent AI will rule the world
Ever since 1984, we have been plagued with visions of a world-wide superintelligence that is out to destroy humanity. But Terminator’s Skynet is just one of many manifestations of this particular AI myth. In fact, this dystopian myth goes back to the earliest days of SciFi. The reality is very different though.
Currently, Google’s DeepMind is arguable the most intelligent and powerful AI. DeepMind is capable of amazing feats, like learning to play the Chinese board game, Go. However, that took it many months to achieve. Likewise, it is capable of outperforming human radiologists at identifying breast cancer. But that was the result of months and months of training.
So, in a few specific cases, AIs can outperform humans at certain tasks. However, this falls far short of even matching human intelligence, let alone reaching the level of superintelligence. This is in large part because AIs have to be designed and created by humans with human intelligence.
I need to understand AI to trust it
As discussed in our recent blog, trust and ethics are increasingly important aspects of AI research. In particular, people are wont to distrust AI, not least because of all the AI myths out there! This has led to a drive for explainable AI. That is machine learning models where we humans can explain exactly why a given result happens.
Arthur C. Clarke famously wrote: “Any sufficiently advanced technology is indistinguishable from magic.” This is one of the main reasons for our lack of trust in AI. As a species, we tend to distrust things we can’t understand.
“any sufficiently advanced technology is indistinguishable from magic”
However, there are countless examples where we are able to overcome this inherent distrust. For instance, almost no one really understands computers, and yet we use them all the time. In fact, this has been true for almost every invention in history. At first, we are mistrustful, but as the technology is more widely adopted, we learn to trust it.
Companies don’t need an AI strategy
Many companies have already embraced AI. However, many are still dubious about the benefits. The problem is, there are all too many examples where AI has failed to deliver the promised benefits. Usually, this happens where someone has unrealistic expectations of what an AI solution can do. Maybe because they believe AIs are more intelligent than they really are. Other times, it’s because people use the wrong AI solution for the job.
Even companies that are all for AI often lack a coherent AI strategy. This means AI projects are often pursued by individual managers with no coordination from above. This has a number of clear drawbacks. For instance, each resulting AI solution tends to operate in isolation. They can’t share data, let alone cooperate to create more complex systems.
In my opinion, every company in the world will benefit to some extent from AI. However, you need a realistic strategy. This requires:
- Realism. Understand the true capabilities of AI. It isn’t a magic bullet. If your business model is fundamentally flawed, you can’t solve that with AI.
- Bravery. AI can only transform a business if you embrace it wholeheartedly. You shouldn’t expect miraculous results from small AI deployments. Instead, you need to find projects where it can make a real difference.
- Trust. You have to win over the trust of your workforce. They will be naturally suspicious, largely because so many people believe the AI myth about job losses.
- Patience. Creating a working AI solution can take many months, even years. This doesn’t sit well in a culture of instant wins and short-termism.
Ultimately, what is really needed is buy-in and leadership from your whole board.
You need data scientists to implement AI
The last myth is that AI is the sole preserve of data scientists. This AI myth is one of the trickier ones, because (like any good conspiracy) it is based on reality. Certainly, creating machine learning models from scratch requires skilled data scientists. Equally, data scientists are best-placed to understand the potential of AI in your business. However, there is an increasing number of platforms that make it easy to create ML models. The best of these can operate without any expert input.
Here at Sonasoft, I have helped drive this trend with our NuGene platform. This is one of the most advanced AI platforms. We specifically designed it to be easy for non-experts to use. We did this in two ways. Firstly, NuGene learns from your data autonomously. Secondly, it incorporates many of the key aspects of robotic process automation. In other words, what it delivers are AI bots that are easy to integrate into your system. As a result, we define NuGene as an AI bot factory.
TRY IT FOR FREE
We are confident that our products will exceed your expectations. We want you to try it for free.