Artificial Intelligence (AI) has attracted the entire world’s attention. The question is, whether we are actually paying attention to the right aspects of it and truly understand it ?
How can the world prepare for it when it is unable to see what the real problem is ?
A number of foreign news websites reported that Facebook decided to discontinue its chat bot because it had “invented its own language” – one that humans do not understand, to be exact. This sparked many discussions overnight, considering that this was not a mere back office update. It fed the existing fear that humans had of Artificial Intelligence (AI), specifically that at some point, humans will no longer be able to control AI, and AI would then take control over the human race.
The truth is the fact that AI had invented its own language isn’t as scary as it sounds !
After the incident, many news publishers and bloggers (e.g. Forbes and blognone) announced that this news strayed far from the facts. Although it is true that chat bots are able to invent or form its own language, Facebook did not discontinue the chat bot option out of fear, but because the way how it had operated was not in line with what it had been programmed to do (which was to communicate with humans in English). In light of this fact, as large media corporation such as QUATZ summarizes, AI was discontinued not because it was too smart – rather because it was too dumb !
So what does the world need to worry about? Do we still need to prepare ourselves for this “newly-invented language” ?
Dhruv Batra, a professor of Georgia Tech and Scientist at Facebook AI Research (FAIR) claims opines that it appears as though there is nothing that we should be concerned about, and that the “synthesized” language is quite normal. He said that “although the thought that AI is able to invent its own language seems alarming and unexpected to people outside the field, in reality, many people who work in AI are aware that this has happened in the past, with reports dating back decades.”
FAIR, where Dhruv is working, also issued a document to certify that this matter was no secret. One of the primary objectives of research work is, in addition to ensuring that AI is able to communicate (talk), to ensure that AI can enter into negotiations, regardless whether among themselves or with humans. This is not relevant to the question of whether or not AI is able to invent its own language. On the contrary, it depends on how meaningful the interaction among themselves and with humans is. It seems as though AI is fast learning and improving its services on its own. Sometimes it is difficult to discern whether the “person” responding is a human being or a robot. Some said that all such changes are in line with the development procedures.
Mr. Rungroj Tancharoen, Chief Executive Officer of Rabbit Digital Group
For Mr. Rungroj Tancharoen, Chief Executive Officer of Rabbit Digital Group, a new-generation digital agency company who is interested in deep tech and constantly stays abreast of AI developments:
“I think that the ability to invent a language is not the primary reason in this case. It’s just one of the things that AI can do. If you go back to read the articles on this topic, you will find that the purpose was to ensure that AI can help negotiate, particularly how AI can make the communication process extremely efficient. The human language is not the most effective one out there. If you ask me whether we should be concerned that the two chat bots are able to communicate with one another and make negotiations, but such communication was not in a language that humans understand, I would respond that no. There is no need to completely fear this development, considering that it is merely a means through which AI communicate with one another most efficiently.”
What most researchers are more concerned about what is called AI’s black box. The black box is a box in which we enter input, which then yields an output. However, we still do not understand why such input yields that particular output, and what method or logic was used in that box in order to arrive at that output.
In the past, studies were conducted on computer algorithms. Regardless of how complex an algorithm may be, programmers were still able to read and understand how they worked. In the case of the black box of AI, on the other hand, this is something which we do not understand, be it the cause or the effect. An output is achievable upon entering in hundreds of thousands or millions of data sets, which the black box then learns on its own based on the data entries. The learning rate is very fast, and, for this reason, the thinking method involved with the black box is extremely complex and humans simply cannot ‘catch up’.
Photo credit: https://cdn.geekwire.com
For example, AlphaGo, an AI technology, defeated its (human) competitors at the world Go Championship. We are not capable of analyzing the detailed approach on how AlphaGo had defeated humans. We ask ourselves, well, what is it that AI was thinking when it had taken a particular move during one of its turns in the game? For this reason, we fear that if the black box continues to work at this pace and is able to resolve issues that are more serious or in-depth in nature, will we still be able to understand and catch up? In a sense, it is like we encounter someone who is far more intelligent than us. We do not know much about his or her personality or what he or she was thinking about, or what those thought processes had involved. This means that we do not know whether the black box had used its intelligence to come up with something that was beyond control. This, in actual fact, is something that should be more of a concern.
“Another factor that is more problematic than the invention of its own language is the fact that it has already begun taking away jobs from humans in the short term (not to mention the long term fear of AI taking over the world). For example, if we look at the industrial revolution, at that time, many people feared that machines would replace humans. In the end, it was found that yes, this was true. However, at the same time, there were reports that the people whose jobs were replaced by AI then changed to work in the service industry. For example, a US worker moved to another employer to work in a fast food restaurant.”
Waves of change occur numerous times. In this case, the difference is that AI is so advanced to the extent that it is capable of completing many tasks that humans can do, e.g. painting, song writing, comedic acting, and more. It got to the point where there was a rumor that if a man tries to hit on a woman at the bar, AI would be able to do so even faster than an actual human being, because their soft skills (or social interaction abilities) are so much better than ours. This is where the difference lies. In the past, we ran away from machinery and turned our focus to service provision, and, as a consequence, the framework of the economy changed accordingly. However, in this case, it appears as though we have run into a road block, because in the end and ultimately speaking, AI will excel and be able to perform even better than human beings in all respects. We will find out where the human race is headed within the next ten years.
It is unlikely that we would be able to see the real consequences of AI’s contribution until it is no longer far into the future. Today, many people still find it difficult to imagine whether that is even possible. Elon Musk and Bill Gates, owners of giant businesses and technologies, as well as Stephen Hawking (a famous physicist), on the other hand, continue to invest significant money and time in order to show the world that AI is a concern and that preparations must be made.
According to Venturebeat (a technologies website), what we should actually be preparing ourselves for are:
1. The rise of drones as a weapon of war
Drones are becoming more widely used in order to reduce the number of injured soldiers. The pricing of these drones is significantly lower than other weapons of war. For example, jets cost USD 100 million, but drones only cost USD 13 million. This is deceptively simple. Conversely, however, the lower price means that anyone can own and use those drones to assassinate others. A simple search on Youtube reveals that drones with automatic rifles and bombs can be used to make a kill from distances of over one kilometer, which will certainly have an impact on the national security in the future.
2. Control of AI
What is concerning is not the invention of a language or AI itself, rather, who controls it. Considering that its power can determine the direction of the world and its very being, if AI falls into the hands of power-hungry people or nations, this may lead to misuse. For this reason, many researchers and influencers in Silicon Valley
came together to establish OpenAI
, a non-profit AI research company, in order to prevent these fears from becoming realities in the future.
3. Growth of AI
Due to the likelihood of the exponential growth of AI, in conjunction with the assumption that it will advance far beyond the intelligence of its creators and be able to survive all eras of change in the world, we thus need to establish an organization or institute that will be able to address this unregulated growth and understand the ethics of AI development, potential limitations, and ensure that security measures are in place in the event of an emergency.
Lastly, we must realize that even though there are numerous concerns regarding AI, we should not entirely close ourselves off from new developments simply because of news that may have caused fear or panic among people. Such news is entirely normal in the case of a new system that is undergoing testing and may be misleading, making us blind to the issues that should be taken into serious consideration and for which we should prepare ourselves.