1. Introduction
2. Artificial Intelligence (AI)
2.1 Semantics (AI and Singularity)
2.2 Hopes and Expectations – Why are we trying to create Artificial Intelligence?
2.3 Strong AI vs. Weak AI
2.4 The Turing-Test and Captchas
2.5 Areas of Research
2.6 Research Approaches
3. Silicon Valley
3.1 General Information
3.2 The importance of Silicon Valley in the history and future of AI
3.3 History
3.4 Silicon Valley as entrepreneurial inspiration
3.5 Practical applications of Artificial Intelligence in Silicon Valley
3.5.1 Tesla Inc. Autopilot: self-driving cars
3.5.2 Apple Inc. – Siri
3.5.3 Open AI
4. Technical and Ethical Limits
4.1 Moravec’s Paradox
4.2 Can machines attain consciousness?
4.3 The social, legal and economic impact an AI could have
4.4 Moral dilemmas – driverless cars
4.5 Getting Lost in the Virtual World
4.6 Dystopian Novels and Movies - QualityLand
5. AI – Chance or Threat?
5.1 Introduction
5.2 Stephen Hawking
5.2.1 General Information
5.2.2 Stephen Hawking’s opinion on AI
5.3 Alan Turing
5.3.1 General Information
5.3.2 Alan Turing’s opinion on AI
6. Conclusion
Glossary
List of Literature
Bibliography
The research paper at hand examines the issue of artificial intelligence (AI) (or: machine intelligence) and its famous breeding ground Silicon Valley. AI could shape our future and determine in which direction humanity will be heading in the next decades or even centuries. Is AI a chance or a threat? To answer that question, it is important to find out how researchers want to create AI and for what reason.
As many sci-fi movies revolve around AI, I have seen many apocalyptic scenarios, still I have never approached the subject from a scientific view and also have never dealt with the topic closer than watching unrealistic Hollywood movies about AI. But how close to reality are those apocalyptic scenarios?
As a matter of fact, AI could make or break our future, therefore I believe it to be very important to be informed at least about some basics of this modern technology.
This research project was inspired by the satire novel “QualityLand” by Marc-Uwe Kling. The novel tells the story of a world dictated by technology and I wanted to see how realistic the scenario of the book could be.
First, this project will explain what AI is, why we are trying to create it, how researchers intend to create artificial intelligence and how you can categorize and classify different forms and types of AI.
Secondly, this paper will give attention to Silicon Valley, one of the most technically advanced places in the world. With its many resources and bright minds, Silicon Valley plays a key role in AI research. This paper will provide a short briefing on Silicon Valley’s history and spirit and then move on to present products developed in Silicon Valley which possess machine intelligence.
Afterwards, it will reflect on problems that might come up and give a brief overview of ethical discussions and technical limits. Moral dilemmas and problems AI could cause will be illuminated and considered.
Lastly, expert’s opinions, such as Stephen Hawking on the question: “AI - chance or threat?” will be elucidated.
The aim of this project is to inform and help people to form a differentiated opinion on the question Stephen Hawking posed. This project also helps me to find out more about this very important topic that I was completely ignorant of before.
Artificial Intelligence is a branch of information technology. It deals with the simulation of intelligent behaviour. To define the term “artificial intelligence” precisely is difficult, because there is no explicit definition of intelligence itself. Nonetheless it is widely used in research and development.
Singularity in the context of AI is the moment when AI’s are intelligent enough to improve and advance themselves without human help. This is seen as a crucial moment regarding AI.
Artificial Intelligence fascinates people. Even though many warn of machines being a big risk and believe a superintelligence will doom humanity, computer scientists keep researching and trying to create artificial intelligence.
AI’s help to deal with the unrestrained data flood generated by sensors and linked devices.1 They can analyse and interpret data.
AI’s can also be useful digital assistants, managing appointments and supporting people in general, enabling them to master everyday life better, easier and raising its quality. They can relieve humans from easy, repetitive tasks.
But AI’s can not only be useful assistants; medical scientists use AI’s to diagnose diseases and to try and find a cure for them. For instance, AI can help with cancer research.2 Other uses of artificial intelligence will be explained further in chapter 3.4 – AI in Silicon Valley.
Summing up, AI’s can be very helpful in performing many tasks, such as analysing data or diagnosing diseases.
In the next chapter, AI’s will be categorized into two different categories.
There are many ways to categorize artificial intelligence.3 One approach is to differentiate between strong and weak AI’s.
A weak or narrow AI is trained for a particular task via pre-set algorithms. Virtual personal assistants such as Apple’s Siri are a weak AI’s.
An AI is classified as strong if it has the cognitive abilities of a human. When presented with an unfamiliar problem, a strong AI would be able to solve it, because it is intelligent enough to do so. A strong AI also has consciousness and emotions. Visions of creating strong AI’s remain futuristic. Nonetheless, strong AI’s make appearances in many movies and books.
To this day, the Turing-Test has a major role in determining whether a machine is intelligent or not. Developed by the British mathematician and AI pioneer Alan Turing, it is a modification of the ‘Imitation Game’. The ‘Imitation Game’ is played with three people, a man, a woman and an interrogator. The interrogator has parallel conversations with both of them. Being in separated rooms, they can neither hear, nor see each other but rather communicate through typed messages. While both dialog partners pretend to be the woman, the interrogator is to determine who is who.
In the modified version, a machine, which tries to act like a human would, takes part in the game and the interrogator is to determine which of his dialog partners is the machine and who the human.
If the machine passes this test and the interrogator can’t identify the machine, the machine is classified as intelligent.
Alan Turing explains this test in his paper “Computing Machinery and intelligence” from 1950. In the paper, he also states:
“I believe that in about fifty years' time it will be possible, to programme [sic!] computers, […] to make them play the imitation game so well that an average interrogator will not have more than 70 per cent chance of making the right identification after five minutes of questioning.”4
But to date no AI has ever passed the Turing-Test.5 And scientists believe, that no AI could pass the test at this time.
A practical application of this method are Captchas. Based on the Turing-Test, Captchas are small pictures with unsorted characters and numbers. Widely used to distinguish ‘Spambots’ from humans, CAPTCHA is an acronym for “Completely Automated Public Turing test to tell Computers and Humans Apart”.
The work on the field of artificial intelligence can generally be broken down into these four subdomains: robotics, computer vision, natural language processing and knowledge engineering.
Robotics is concerned with designing, constructing, operating and using robots which are able to autonomously perform physical labour.6 Engineers try to automate dangerous operations such as welding. The challenge here lies in the conjunction of different techniques like sensor systems, data analysis and action planning and –executing.7
Computer Vision tries to teach computers to “see”. Here the goal for the machines is to be able to interpret and analyse visual images with the help of a special form of neuronal networks: convolutional neural networks (CNN). Pioneered by Kunihiko Fukushima, Geoff Hinton and Yann LeCun it consists of billions of highly connected neurons. Neuron-like units (nodes) are organized in a hierarchical order. The structure is inspired by the human brain. The nodes take input from other nodes and send output to others.8
Natural Language Processing (NLP) is about trying to understand the mechanics languages, like its structure, handling and application. A machine should be usable for human-machine interaction.
In knowledge engineering, the main focus is on developing assistant- and expert systems. Every expert system has a knowledge base with facts and rules, an inference machine – something like a conclusion program – which checks if a statement is compatible with the knowledge base, a dialogue module or the user interface, so a person without background knowledge can use the machine, a trace module showing the history of argumentation and a knowledge changing element, where facts and rules can be added and edited. An example for an expert system would be a system for diagnosing diseases.9
Generalized, there are two different research approaches to creating AI, the first one being the Physical Symbol System Hypothesis (PSSH). This approach was very popular in the 1970’s and 1980’s. Supporters of this system deem it to be necessary for making a machine act intelligent to understand the process a brain undergoes to find a solution for a problem. This hypothesis assumes that symbols form the basis for intelligent behaviour. Human cognitive skills are perceived as symbol manipulation. Symbol manipulation stands for the processes that are responsible for cognitive skills. The insights into cognitive processes of the brains they gain this way are to be translated into machine algorithms. Admittedly, nowadays the PSSH is not nearly as popular as it used to be. However, nobody can say for certain if symbols actually form the basis for intelligent behaviour. These days there are other options for creating AI, regardless symbolic AI is not unneeded, it just does not cover all aspects.
Another approach is machine learning10 . Machine learning is not contradictory to symbolic AI. Yet, while a symbolic AI only has a learning phase at the very start, in machine learning the importance and benefit of constantly learning are recognized. Learning does not only mean gathering and saving information but rather reprocessing information so it can be of use later on. A special approach in machine learning are artificial neural networks. Neural networks are computer programs where the organization of neurons is inspired by a human brain. In so-called computational neuroscience, scientists try to understand and simulate the functionality of the brain. The issue here is, how little we know about the operating system of the brain. Moreover, these networks indeed are very fast, the applicability is restricted to sensory tasks though.
In the southern portion of the San Francisco Bay area in the northern part of the U.S. state of California lies the USA’s – and the world’s – high-tech heartland. It is internationally known as Silicon Valley, the centre of engineering breakthroughs and the country’s top centre for high-tech employment.11
As a dominant area for technology manufacturing it is not only a magnet for hundreds of software start-ups but also home of many of the world’s largest high-tech corporations. Apple, HP (Hewlett Packard), EA (Electronic Arts), Google, Netflix, Facebook, Tesla Motors, Intel, eBay and Yahoo! are all based in Silicon Valley, to only name a few.
Designed to be fully functioning communities, the campuses of the local companies allow employees to shop, eat, go to the doctor and even have their hair done without leaving work.
Once defined by its factories, Silicon Valley now stands synonymous for product design and interactive digital media. However, Silicon Valley is not only a physical place but also a state of mind. For outsiders, it might be hard to grasp the spirit of Silicon Valley whereas employees have a very different view on the world.
Employees share a strong believe that they can change the world. From their point of view, technology is the most important field of research. The government is considered slow and people from the valley often think they are faster and more efficient than the government.
“At places like Facebook, it was felt that making the world a more open and connected place could do far more good than working on any charitable cause.”12
As this quote shows, many employees believe their work to be more important than the governments, explaining their dedication to their jobs.
In recent years, Silicon Valley also demonstrated its ability to reinvent itself relentlessly.
This process can also be called “creative destruction” and it helps Silicon Valley to remain dominant.
All in all, a lot of the most advanced technology has emerged from Silicon Valley and it continues to thrive.
As mentioned above, Silicon Valley is technically advanced and very many innovations come from Silicon Valley.
Thus, Silicon Valley is and has always been very important for the development of artificial intelligence. The first predecessor of a computer was developed by the founders of the Silicon Valley based firm HP (this will be further explained in chapter 3.3 – History). As Silicon Valley is so successful, it has many resources and the innovation promotional atmosphere is just right for smart minds trying to create AI. The goal of Silicon Valley firms is to revolutionize – to make progress. Today, the American Association of Artificial Intelligence (AAAI), based in Silicon Valley, hosts conferences on AI, to “ promote[s] research in AI and scientific interchange among AI researchers, practitioners, and scientists and engineers in related disciplines”13.
With its funding and urge to innovate, Silicon Valley will play a big role in the future of AI, as it did in the past.
The beginnings of Silicon Valley lie with two graduates of the renowned Stanford University in Palo Alto near San Francisco, William Hewlett and David Packard. In a garage, they designed the first predecessor of a modern computer. At this location, they founded the now extremely successful firm Hewlett-Packard.
Silicon Valley got its name from Electronic News reporter Dan Hoefler who first used it in a report about Silicon Valley and the name stuck.
The centre of the American electronic and computer industry first began attracting national attention in the 1970’s during the computer revolution. At that time, Silicon Valley had grown to around 15 companies.
“During recent years [sic!] the San Francisco Bay Area developed rather suddenly into one of the major centers of electronics research and industry in the United States. To those who knew the background it seemed a natural evolution in a region that has been the scene of radio and electronics pioneering since early in the century.
-Frederick Terman, from preface in Morgan 1967”14
But, as Frederick Terman - who by many is considered the father of Silicon Valley – states here, Mr Terman didn’t actually found Silicon Valley. Rather he can be considered as a catalyst and booster in an already prepared environment.3
An important aspect of Silicon Valley’s success are their disruptive business models which help companies to constantly evolve. To secure its economic leading position Silicon Valley is submitted to constant change.15
Kōnosuke Matsushita, founder of the Japanese company Panasonic, understood, how a foresighted company policy should be like. His management methods were the same as the ones being used by companies based in Silicon Valley.
“Beginning today, this far-reaching dream, this sacred calling, will be our ideal and our mission, and its fulfillment the responsibility of each one of us. […] The most important thing is that we enjoy happiness to the fullest in our own lives and at the same time strive for the benefit of the generations that are to follow”16
So-called “social architects” from Silicon Valley developed a very interesting method to keep employees engaged and to lead a company to success: Future Management. What has to be done to be able to achieve the future we want?
Scientists in Silicon Valley want to change the world. To do that they need to think outside of the box, they need to get rid of borders in their heads, forget structures and processes they are used to, to enable them to get a new perspective on things.
The company needs to build a so-called “thinking space” for them where they can think about anything they want. Illegitimate thoughts maybe even illegal thoughts are allowed here, for the sake of progress. It is all about pushing and changing a company’s limits to destroy normality.
Managers in Silicon Valley found the connection between the future and the present. The things you are focussing on now are the ones which will be important in the future.
Right now, in this very moment, you can influence and perhaps even control the future.
Fictional leaps in time and thought experiments are used to find out what is necessary to achieve the future you want.
“Destroy-your-business” scenarios were thought experiments of Jack Welch, former General Electric CEO called. Even though General Electric is not based in Silicon Valley and Jack Welch himself does not have any connections to it either, he performed Silicon Valley’s management methods and was named “Manager of the Century” by Fortune magazine in 199917.
Jack Welch would let a group of people put themselves in the position of their biggest competitor. This group of people would try to think of the easiest and most efficient way to destroy GE. Helping to identify weak points and illustrating chances GE had, these scenarios helped GE to reinvent themselves again. This is the spirit that drives the Silicon Valley. Thus, GE could understand other points of view and establish alternatives. This kind of thinking is called “cognitive stretching”.
Another method is called Backcasting. Here, a group of people thinks of a utopian world and tries to deduce things needed to be done to obtain this world.
To sum up, the future based decision-making in Silicon Valley is independent from what applies today and it stands for another view of the world. Inflexible rules, known structures and processes are being avoided. This disruptive business model has helped Silicon Valley to become so successful.
Tesla Inc. is rumoured to be working on its own AI car chip for autonomous driving. Hiring Jim Keller, a well-known chip designer who previously worked for Apple Inc. as “Vice-President of Autopilot Hardware Engineering” and poaching a team of chip architects and executives from AMD (Advanced Micro Devices; chipmaker company)18 fuel the rumour. Currently, Tesla vehicles use Nvidia graphics processing units19 but many sources suggest that Tesla is working on its own AI car chip with the help of AMD.20 Elon Musk, Teslas CEO also promised a completely automated Tesla by the end of 2018.21
Even though it is controversial, Apple’s Siri could be considered a weak AI, more accurately an assistant system but also a system able to perform natural language processing (see: chapter 2.5 – Subdomains). Aired in 2011 it is supposed to facilitate the handling. It is controlled through speech. To improve Siri’s voice, Siri’s engineers used a machine learning algorithms which is also used to help the machine to better understand what the user wants. A recent addition to Siri is the HomePod, a Siri operated speaker system.22
[...]
1 cp. (Mainzer, 2016)
2 (Malirsch, 2017)
3 cp. (Rouse, 2016)
4 (Turing, 1950)
5 cp. (Huber, 2009)
6 cp. (Kaplan, 2017)
7 cp. (Görz, et al., 2003)
8 cp. (Li, 2015)
9 cp. (Zöller-Greer, 2007)
10 cp. (Flach, 2012)
11 cp. (Pimentel, 2008)
12 (Packer, 2013)
13 (A (Very) Brief History of Artificial Intelligence, 2006)
14 (Sturgeon, 2000)
15 (Müller-Friemauth & Kühn, 2016)
16 (Kotter, 1997)
17 cp. (Lee, 2015)
18 cp. (Lambert, 2016)
19 cp. (Owens, 2017), (Novet, 2017)
20 cp. (Ray, 2017)
21 cp. (Wales, 2018)
22 cp. (Wobker, 2017), (Shankland, 2017), (Pierce, 2017), (CBS News, 2017)
Der GRIN Verlag hat sich seit 1998 auf die Veröffentlichung akademischer eBooks und Bücher spezialisiert. Der GRIN Verlag steht damit als erstes Unternehmen für User Generated Quality Content. Die Verlagsseiten GRIN.com, Hausarbeiten.de und Diplomarbeiten24 bieten für Hochschullehrer, Absolventen und Studenten die ideale Plattform, wissenschaftliche Texte wie Hausarbeiten, Referate, Bachelorarbeiten, Masterarbeiten, Diplomarbeiten, Dissertationen und wissenschaftliche Aufsätze einem breiten Publikum zu präsentieren.
Kostenfreie Veröffentlichung: Hausarbeit, Bachelorarbeit, Diplomarbeit, Dissertation, Masterarbeit, Interpretation oder Referat jetzt veröffentlichen!
Kommentare