Can Artificial Intelligence Algorithms Solve All Your Business Problems?

Cryptocurrencies are a good example. The first cryptocurrency run on a decentralized blockchain was Bitcoin and it launched thirteen years ago. Once it got more widely adopted other cryptocurrencies started to emerge on the market. Some of them might have a relevant use case behind their inception, but others not so much. Nowadays there are over 10,000 different cryptocurrencies on the market, and no one believes that they all solve a specific use case.

Similarly, a big hype also appeared around artificial intelligence (AI) and machine learning (ML). There are influencers and companies who claim to be solving all and every possible issue with the use of AI. (Un)fortunately, AI and ML are tools (very efficient, but not more than that), which at least for now, and for the near future, can solve only a handful of challenges.

Here is why. When we talk about “what is artificial intelligence?” we need to differentiate between two fundamentally different forms. These are defined as general artificial intelligence and narrow artificial intelligence. These two are lightyears apart from each other. Whereas narrow AI is created to solve one specific problem e.g., recognize the face of a specific person, general AI is a theoretical concept where machines can think for themselves, just like humans, deciding which problems to solve, how, and with which tools, etc.

Narrow artificial intelligence

Narrow AI is used for repetitive, narrowly defined, and structured tasks involving vast amounts of data. In these conditions the narrow AI thrives and can easily outperform human abilities. Hence, its popularity, among (not just) the big tech companies, to use it to provide a better service to customers at a cheaper price.

Some of the real-world examples of narrow AI are:

• Recommender systems: Amazon, Spotify, Netflix, and other platforms leverage the vast amounts of data they have about our behavior. They use it to recommend products, songs, or television series, which we will find most interesting. They use this data to create a profile based on our behavior on their site, then they will match our profile to the attributes of products or other users. We, at Simplity, create similar recommender systems for our e-commerce clients.

• Facial recognition: Facial recognition systems are being used in a number of different ways e.g., to provide authentication, recommend the right people to be tagged in photos and videos, or index images. Nowadays it is also being used for security purposes, which provokes a lot of controversies and disputes.

• Disease detection: Studies have found, that machine learning can detect symptoms of specific diseases faster and more accurately than healthcare professionals due to having access to larger datasets and faster processing. In the future, medicine could leverage these capabilities and enable healthcare professionals to focus more on primary care.

• Search systems: Google, Bing, as well as other search engines bound to a single webpage (e.g., search on Amazon) leverage narrow AI to better understand the user intent, autofill users’ queries, and provide the most relevant results.

• Text autofill: As I write this article, Microsoft Word leverages narrow AI to autofill the words and sentences, which nicely complement my statements.

There are multiple other working use cases of narrow AI but all of them work the same way. First, you need to have a lot of data from which the narrow AI algorithm can learn. Second, once the algorithm has learned from the historical data, you let it make predictions based on the common attributes. Third, these predictions continually improve over time as the algorithm makes new connections from the new data.

General artificial intelligence

We have seen considerable progress in narrow AI over the last two decades. General AI is where the researchers are heading. Whereas narrow AI can run single, automated, and repetitive tasks, the general AI should be able to think and reason like humans. As you can imagine, the difference between these two types of AI is huge. Narrow AI is relatively simple as the algorithm learns only from a closed set of data and predicts future outcomes based on probability.

General AI models, on the other hand, should mimic the way our incredibly complex brains work. Since scientists still have very little understanding of how our brains work, we can be nowhere near mimicking them, in order to create the general AI.

Conclusion

Narrow AI can be, and already is widely, used by businesses to improve the customer experience because it exceeds the human abilities in simple, repetitive, everyday tasks. However, it is just a tool which can process data faster and in larger quantities than humans, which makes it applicable only to a handful of business challenges. General AI, on the other hand, should be able to think similarly to human beings and should have attributes such as common sense, background knowledge, transfer learning, or abstraction. Nevertheless, nobody really knows when the general AI will be achieved and if ever. Therefore, for now, businesses can only leverage the capabilities of narrow AI.

If you need help understanding what processes and problems AI can really help your business with, we would be happy to discuss them with you.