What is artificial intelligence?
Artificial intelligence (in English: AI – Artificial Intelligence) is a field in computer science that talks about the ability to write code for a computer program, which will display the ability of a quality and level that characterizes the quality and level of that ability for humans. Writing a program that will play chess against you is not an example of artificial intelligence. To write a program that will play checkmate against you, beat you, and can even beat skilled and experienced players – this is already artificial intelligence. Artificial intelligence is created as soon as the computer manages to respond to events in a way that resembles human intelligence.
It is customary to perform the Turing test against the software application to determine whether it is artificial intelligence. Alan Turing first defined this test in 1950. As part of the test, an examiner is in a closed room, and he conducts the interaction twice – once with the software and once with Ben, a real person. If at the end of it, the examiner can say which of the two was the human who responded to him and which was the computer program, the software has not passed the Turing test. If the examiner at the end announces that it was difficult for him to understand in which of the two cases it was the software, and in which of the two cases it was a person, or if he guesses who is the person and who is the software incorrectly – the Turing test is passed.
If the process is repeated and 30% or more of the cases pass the Turing test, that software will be considered artificial intelligence for all intents and purposes according to the Turing test.
The Holy Grail of Artificial Intelligence: Human Conversation
There are areas where we are used to perfect artificial intelligence and even perfect since the dawn of time, there are cases where artificial intelligence is impressive but possible, and there is the holy grail of artificial intelligence – human conversation.
An example of a field where artificial intelligence has put the ability of humans in a small pocket for many years: is mathematical calculations. Any pocket calculator knows how to perform complex and complicated mathematical calculations much better, faster, more accurate, without mistakes at all, and at a much higher speed than any human being. Artificial intelligence is stable and strong in this field, and it even exhibits higher abilities than human abilities in the field.
Artificial intelligence is very impressive but possible if we refer to the checkmate example that we brought before. Chess software can play against the best human players in the chess world and beat them. the truth? Even in this example, the computer has an advantage – it usually plays much faster than the human player.
From here, we move on to the holy grail of artificial intelligence: human conversation. Nowadays, even the world’s leading software companies, including the giant’s Facebook and Google, are still decades of development (at least) away from enabling software that knows how to talk to us humanly. The two companies mentioned, Facebook and Google, definitely try and show impressive purpose displays, so it seems that the direction is completely correct.
Facebook has its new chatbot: Blender, which, when its development is finished, will allow business owners to place an automatic response on their Facebook business page, an automatic response that will converse in a completely human manner with the potential customer who is contacting the business. Google has “Mina,” the personal assistant that will have a conversation with you completely humanly, and there is Google Duplex. Some conversations with Google’s assistant “Mina” are very impressive. Google presents examples of responses and conversations with Mina, among which there are cases where Mina responds with humor, wit, and sharpness and answers difficult and complex questions with logic and precision. I .did not write these responses for her in advance; she puts together sentences in a computerized way, in a precise way, and sometimes even as the saying goes – humorously.
Google Duplex is a service in advanced development that has already been released to the general public (only in English at this time) by Google that will be able to reserve a place for you in a business through a computerized voice call. Suppose you want to make a reservation at a restaurant or hair salon, and the same hair salon or restaurant does not have a mechanism for booking a place online. In that case, you can order Google Duplex to make the reservation call for you.
Just tell Duplex for which place (or tell it to locate a recommended place nearby), the date and time you want to arrive, and if it is a restaurant. How many places do you want to book if you intend to come with people? Google presented two examples of 2 conversations (these are voice phone calls) that Duplex conducted, one with a restaurant and the other with a barbershop. The Duplex had conversations that sounded completely human. The other party did not know and had no chance to guess that it was a computer program and not a human being talking to him (Duplex passed the Turing test in both cases by a large margin). In an example conversation in front of the restaurant, a girl answered on the other end of the line with a heavy Chinese or Japanese accent, which was sometimes difficult to understand what she was saying (in English). She also had difficulty understanding English since it is not her mother tongue. Duplex managed the conversation despite the difficulties, rephrased more clearly when he was not understood, and understood what she was trying to say even if she spoke with a heavy accent and grammatical errors.
The girl from the restaurant told Duplex that they don’t reserve a place for four. Only from five places do they allow a reservation in advance. Duplex expanded to do and asked how the availability was and if there was anything to worry about the arrival of four people at the time. Is there a chance that there will be no room? The girl from the restaurant calmed down and said that it was not a busy time and they could come freely without a reservation. Duplex responded wonderfully: “Oh, okay, I got it, great, thank you very much”! Very impressive. Duplex speaks in a completely human voice, adding “aha” and “umm” just like a human to sound more human.
Another company that develops interesting artificial intelligence services: is IBM has developed a computer program that can conduct a debate on a certain topic. The software expresses arguments for or against (defined in advance), listens to the other side’s response, negates his arguments, and adds arguments to strengthen his side.
How does artificial intelligence affect our lives?
Artificial intelligence these days is already affecting our lives, even in ways we still don’t understand. It will continue to affect our lives in crazy ways that will sound like a fantasy.
Here are some examples:
The facial recognition in your cell phone is artificial intelligence. The ability to identify who was standing in front of you was the property of humans only. No software knew how to properly identify if there is someone in front of it who is approved for entry (or maybe you show the software a simple picture of that person?). In recent years, software (combined with appropriate hardware – sensors that map the face in three dimensions) knows how to recognize when a person is standing in front of her with permission to enter and when not, and when they are trying to trick her with a photo of the person with permission.
Understanding speech and language and translation
an example of artificial intelligence whose part is still immature. Today, through the Google Translate application, it is possible to conduct a dialogue in two different languages in a way that is certainly worthy of praise. I will speak in Hebrew, what I said will be said again through a speaker in English, the other party will answer me in English, and his response will be said through the speaker immediately in Hebrew. This thing already works and works quite well – but we all know the shortcomings and limited capabilities of the translation services (even the most advanced ones from Google). Sometimes the translation comes out crooked. In a small conversation, Google is expected to translate accurately. Still, when you put an entire academic article into the translation engine (even if it is not academic) – there will be many mistakes in the translation. The speech understanding side works quite well, and the translation side – is less so.
What will be possible when the technology matures?
Communication restrictions worldwide will be abolished as if there were none. You can enter an airport in the Netherlands using glasses or your cell phone camera. . You will immediately translate all the signs around you into Hebrew. You can ask one of the employees at the airport in Hebrew which way your gate is, and he will answer you in Dutch – and you will both understand each other well through an instant translation. A professor from Russia will give a lecture at a university in the United States in Russian, and after a delay of a few seconds -. Will hear his words through a loudspeaker in English, and all the students will understand him. You will be able to read a book originally written in a language that you do not understand in your native language immediately; everything will become accessible and understandable without language and communication limitations.
Artificial intelligence is a critical transition point for computer understanding of language and expressive skills.
This is a point that, when fully blown – artificial intelligence will rush forward rapidly. Language is the embodiment of the expression of (human) intelligence, which distinguishes us from animals. To use artificial intelligence freely, we need to get to a point where a conversation with a computer sounds exactly like a conversation with a mortgage advisor at a bank.
I present questions and receive precise answers that correspond to the questions. The whole issue of deciphering the language and fully understanding what is being said is still lacking. Google and Facebook’s developments are getting close but fail to correctly decipher what is meant in 100% of the cases. This, unlike humans, who, as long as you talk to them in the language they speak, will almost always understand what you say, or at most, they will ask you again what you meant and understand you after you repeat your words in different wording.
How will artificial intelligence affect our lives in the future?
Some things I will tell you about will sound like fantasy. Some are closer, and some are less.
Imagine a world where you can ask your cell phone, in a free voice language, for anything you could ask a human being to do for you. Your cell phone will become a personal assistant for everything and anything.
“Hey Pelephone, I think you need to go shopping, make an order.” The cell phone will ask you what you want to order, which is simple and easy to implement, but what happens if you give it a less predictable answer?
“I really don’t know, I don’t have time to check, just order what you think is missing.”
The cell phone will understand and decipher what you said, run to your order history, see when you last ordered tomatoes and cucumbers and what quantities, try to figure out what is likely to be missing, call the supermarket, and place an order (with a human voice, of course). The next thing that will involve you is the delivery person knocking on the door.
You can tell him to call Anat and explain to her that you have a problem with the meeting you scheduled for tomorrow at noon and that he will try to reschedule. The cell phone will call (in the background) Anat, identify herself as your assistant, tell her about the problem, and try to reschedule with her while using your calendar. He will return a reply that everything is done. At this time, I am in the Ramat Gan area until a little before. Try to change the location with her.” Your cell phone will call Anat again, explain the new matter to her, and settle the matter with her. Fifteen minutes in which your cell phone did things for you, and you freed yourself up for other new things and interesting.
We have already told you that Google’s “Mina” artificial intelligence knows how to respond with humor, to make up random jokes (there are some that even really made me laugh, jokes that the engine made up in real-time and was not written in advance) and to be witty and sharp as a razor. Artificial intelligence can understand humor, let alone produce humor, which is amazing.
Can you imagine a world where a robot will appear among the performers at a stand-up evening? That this robot will make the audience laugh with sharp and witty humor and – will also tell new jokes that it invents and improvises on the spot in real-time, every time it “appears” in front of an audience – a new set of jokes? Can you imagine a situation where the same robot addresses the audience, conducts a dialogue with them, and improvises jokes with them (and about the audience), as many stand-up artists do? This robot endangers the livelihood of stand-ups everywhere. True, it’s not around the corner, but it’s already in its infancy, and this robot will be very funny, like the best stand-up artists, no less.
Let’s sail with the fantasies even more: referees in soccer matches, news broadcasters, doctors in hospitals, all could be replaced by artificial intelligence. Can you imagine a situation where a robot makes medical decisions in real-time, does it better, more correctly and more accurately than a doctor, performs open heart surgery on a patient, and does it as best as possible, better than any other doctor? Crazy.