Artificial intelligence has been a goal since the beginning of computers. Why is it that after over 50 years of work there are still no systems out there that can be said to be intelligent? Computer power has increased by massive amounts, millions has been spent on research but still no major results. This article talks about the current batch of chatbots and where AI should be going.
I should state at this point that I believe intelligence is a set of processes in the brain that is perfectly possible to emulate in software so that a true artificial mind can be produced. I have been interested in Artificial intelligence almost since I first got into computers over 30 years ago. The promise of making computers smart, being able to talk to them seemed like a tantalizingly close dream. How hard could it be? The fame and fortune that would follow the person who first worked it out would be epic. I was hooked.
30 years later and I am still interested but the dream is just as far away now as it was then. What happened?
In a nutshell, nothing much.
Researchers worked on the problem, systems were built but no real results, no systems then or now produce any sign of real intelligence. There are a number of systems (so called chatbots) which are designed to look like intelligence. I guess the most famous is the chatbot called ELIZA which was designed to act like a computer psychologist. It was simple and had a stock set of answers to certain keywords typed by the user. Obviously the results were patchy, it fooled some people but very quickly showed up as just a toy. There have been a number of systems since then that have tried to continue this line of work. I have designed a few myself, they scan the user input to try to get a sense of what the user is saying and then use a set of prepared answers to give the impression of a real person giving real answers. These systems have real limitations and are still no where near what is needed.
There are a number of other systems based on neural nets that are designed to emulate how the human brain works (a series of cells interlinked). Since they are designed to work in similar ways to the way the brain works it is a valid approach. Again after decades of work only a few useful systems have been produced but they are not even close to being intelligent. This approach is similar to trying to work out how a computer works by studying a transistor. It may get some results eventually but it is going to be very difficult.
So why hasn’t there been progression over the years like you get in fields of, study? The answer may seem obvious but it is just too complex. Nobody has figured out how the brain works. If you have billions of cells in your brain each connected to tens of thousands of others, it become impossible to figure out what is going on. Add to this the difficulty in direct experiment on living brains and it’s not hard to see why we are still getting nowhere.
So what do we do? How can we progress after all this time. It is my assumption that if you mimic the way people behave then we can build intelligent systems, we should copy the way our minds work on a system level not on the level of the neuron. If we study what people do, what drives them then we will have an understanding of what is needed to make a system look smart. Stop trying to mimic HOW the brain works, instead mimic what it DOES. Once we can understand in detail what the brain does we can write software that does the same.
So where do we start?
Computers know nothing about the real world. People take years to develop an understanding of the world; this “assumed knowledge” is taken for granted but is a vital part of what makes us what we are. Even a child knows a bird can fly, flames are hot and knives are sharp and can hurt. Computer systems know nothing about any of this, they have no assumed knowledge.
If a computer system is to be intelligent it must have assumed knowledge, it must either have ways to accumulate this knowledge (camera’s and microphones) or we must give the knowledge to the computer. Chatbots or any AI system must have “assumed knowledge”. Ideally a system should learn this assumed knowledge but the technology to do this is another whole subject on its own. So to produce a system in a reasonable time we must give the knowledge to the system at the start.
Knowledge is not everything. Knowing stuff does not make you intelligent, using those facts to your advantage is a sign of intelligence. Assessing advantage is the next key to intelligence. Even simple creatures discover techniques and new methods to get food, fight enemies and stay warm, they assess what happens and form opinions on those events designed to give advantage. Any AI system must be able to assess advantage to it and therefore must have goals. You cannot access an advantage if you don’t know what you want/need.
Mood. All intelligent systems have moods (emotion), sometimes they are angry, happy, sad, violent, and passive. These mood varieties are designed to aid the individual to survive but they also have a direct effect on intelligence, the mood you are in directly effects the way you think, work things out and assess advantage.
In summary all intelligent systems must have:
– Priority (Goals) – Advantage Assessment – Moods (Emotion) + – Assumed Knowledge
Looking to create a chatbot Singapore for your website, Facebook or messenger chatbot? Check out this ai chatbot company that can help you to create Singapore chat app & Facebook chat bot