Artificial superintelligence: The next big evolutionary leap

• Read original article here

According to Gartner, Artificial Intelligence software revenue is forecast to reach $62.5 billion by 2022. The AI software market is growing in every aspect, and executives push their engineering to add AI capabilities to almost every product or service to increase productivity and accelerate innovation. Nowadays, both startups and enterprises are investing heavily to offer their customers solutions with a combination of AI capabilities with AIOps tools to resolve issues before they occur, AI Security to identify and protect customers against cyber security attacks without human innervation, and Chatbots are smarter with AI and can learn from experience; this list is growing each year.

Typically, we use one type of AI called Artificial narrow intelligence that can simulate human intelligence to perform limited tasks in one field, starting with machine learning and deep learning; NLP is another example of AI narrow intelligence. The most common examples of tools and solutions that are using these AI technologies are Apple Siri, IBM Watson, Amazon Alexa, and Tesla self-driving cars.

When it comes to AI, there are three types: artificial narrow intelligence (ANI), artificial general intelligence (AGI), and artificial superintelligence. Artificial narrow intelligence has a narrow range of abilities as it can only perform one task at any time, therefore it is dubbed as weak AI. Then comes artificial general intelligence (AGI) which has the same level of human capabilities to perform tasks, and lastly artificial superintelligence which is software that can exceed human intelligence in every aspect. Artificial superintelligence (ASI) is a type of artificial intelligence with much greater intellectual capabilities than humans. The machines and systems using ASI will be much smarter and perform better tasks than humans in every field or aspect, such as social skills, creativity, and wisdom. Moreover, the artificial superintelligence can perform and execute multiple tasks simultaneously, improve its code, teach itself new abilities and use much better ways to utilize its own compute resources. The day that machines will surpass humans in general intelligence is around the corner– most likely, it will be developed in the next decade.

We already see many AI applications that can make decisions autonomously without any human intervention, such as autonomous databases or autonomous cars. 10 years ago, that may have sounded like a science fiction movie, but today we are developing almost every system or application to work autonomously. We all want to save time, be more productive, and eliminate human effort and manual work as much as we can. In the next decade, machines and applications intelligence will be superior to humans in any field–we could use these machines to find a cure for cancer, develop a new vaccine in just a few days, or be used by criminals and terrorists alike to execute smart attacks like the next fire sale cyber-attack. For instance, as Dr. Noga Yerushalmi, investment director at M Ventures and member of the investment committee of AION labs stated, “The use of AI to try and solve challenges in biomedical research is not new but has become increasingly the focus of new research and novel technologies with the accumulation of more data in this field. Right now, the cost and timeline to develop drugs and get them to the market are only growing, which is bad news for patients, Pharma companies, and investors. I believe AI solutions can be the game changers in this market and enable a smarter drug development process and better precision medicine to help humanity.”

Today, engineers and data scientists work with safe AI because it is human-built and is therefore on a human level. When it comes to ASI, it will be hard to control, but the company that will successfully develop it has the potential to dominate the world.

When will ASI enter our lives? There is an ongoing debate as to when. Some believe that it will happen in the near future. Others, like Aya Soffer, Vice President, AI Technologies, and Director IBM Research, think it is much further away. She stated that “We are not close. In the foreseeable future, it will be artificial intelligence helping humans make data-backed good decisions”. But on the other side of the spectrum, we see people who already believe it exists.

Written by Ido Vapner, CTO and Chief Architect at Kyndryl