Artificial intelligence has been around since 1956 when the term was first coined. Those in the industry know that there has been previous hype and then disillusionment around AI. The period of decline of interest in AI is known in the industry as an AI winter, and has happened twice before. An AI winter is a point at which research, investment, and funding for AI goes into a period of decline. it is hard to get funding for research or other projects that have to do with artificial intelligence and talent and companies focus their efforts elsewhere.
Today, there is a lot of hype surrounding artificial intelligence, but is AI around to stay or will it see its period of interest peak and wane as it has in the past? In order to answer this question, it’s important to learn about AI’s beginnings and past. Are we headed to another AI winter?
The First Wave of AI Adoption, and the First AI Winter
Since the invention of computers, developers have been working to build smarter computers. The term AI was actually coined in 1956 at a Dartmouth convention. Then the first major wave of AI interest and investment occurred from the early 1950s through the early 1970s. There was a lot of interest and funding from government, academic, and military sources which produced some of the earliest and most impressive advancements in AI.
But all this advancement and development was put on hold in the mid-1970s. People started to feel more skeptical about AI’s promises, and the inevitable pullback happened when AI’s promises failed to deliver. This combined with the Lighthill report precipitated the first AI Winter. There are two major reasons for this demise. People had big plans about what AI could do and problems it would solve, but the technology failed to deliver on these grand plans. There was also a lack of diversity in funding sources so when the government and military decided to pull back investment in AI, most of the funding dried up.
The Second Wave of AI Adoption, and the Second AI Winter
Interest in AI research was rekindled in the mid 1980’s with the development of expert systems and reinvigorated research and investment in AI. Corporations started massively investing in desktop computers on knowledge workers desks, and expert systems helped to connect the emerging power of desktop computers and cheap servers to do the work that had previously been assigned to expensive mainframes. Expert systems helped automate and simplify decision-making for many industries. And this wave the funding came from businesses as well as the government, allowing for a little more diversity in funding.
At the same time, as expert systems became increasingly hyped as people wanted to do increasingly more complicated things with them, the more problems that these AI adherents ran into. Expert systems were very dependent on data, and storage was still expensive in the 1980s. Storage and data were a few of the problems that expert systems came across in terms of technological problems. Corporations also needed to develop their own data and decision flows, but faced limitations to what they could really do with this data. In the 1980’s there wasn’t an Internet-based cloud to store data on like we can today, nor access to almost unlimited compute. It was also harder to transmit large quantities of data from one area to another. This meant that expert systems couldn’t communicate with systems outside of the company that owned them and the data they needed to progress further wasn’t available.
These limitations along with diminished appetite for AI lead to another AI winter. In conjunction with expert systems, companies also started to develop software programs to handle various business tasks. Expert systems developed a reputation of being too brittle, depending on specific inputs to get desired outputs and businesses became wary of AI projects when they saw that there were cheaper alternatives that were just good enough to do what they wanted.
AI is hot again, but for how long
For about a decade or so, we have been in a period of a real blossoming of interest and investment in AI, an AI “summer” if you will. The innovation of AI algorithms combined with the availability and experience of working with big data is one of the biggest reasons that artificial intelligence has been able to leave hibernation. We now have almost limitless storage and have been able to successfully manage and handle large amounts of data. The development of deep learning is another reason that we have come out of the AI winter. Also investment is now quite diverse coming from enterprises, governments, academics, and venture capital.
Additionally, dozens of countries have acknowledged that AI is going to be so important for their citizens and growth of their economies that we now have a country level investment and strategies around AI. Furthermore, AI is becoming more a part of everyday life. Today, AI is being used all around the world to accomplish any number of tasks. We have put AI into cars, phones, advanced bots, and other technology that we use every day. It’s not uncommon to interact with AI daily whether that’s through virtual assistants, hyper personalized offerings and recommendations, or better movie suggestions on Netflix. Artificial intelligence now is in so many pieces of technology that we interact with every day that we almost forget that it is there. Sometimes bots are so good that we can’t tell they aren’t humans. We use AI to park our cars, and much more.
However, with all this investment, interest, and funding in AI are we headed to another AI winter? Are we once again overpromising and under delivering on what AI is capable of? Are we going to be disappointed with the limitations of driverless vehicles, natural language processing, and AI-powered predictive analytics? Will investors start seeing more snake oil from AI vendors than real-world implementations?
While there certainly will be a lot of missed expectations for AI, this time around, our expectation and hope is that the next AI winter will never come. Another AI winter is always possible if we have a repeat of past circumstances such as over promising and under delivering what an AI system could do. Some experts are also concerned about the lack of research on general AI, we still haven’t figured out common sense reasoning, and that a lot of the funded research is too application specific. AI is too integrated in our daily lives for research, investment, and use to stop all together, and so it’s possible while there might be a cooling off of research and investment, as there always inevitably is with technology waves, we are hopeful this time we don’t go into hibernation.