Category Archives: Machine Learning

Machine Learning Executive Summary

It is essential to understand what exactly machine learning is and how it works in order to understand the affects it has and will have on society. In basic terms, machine learning is a computer program that can learn and adapt to new data without human interference. It is a field of artificial intelligence that keeps a computer’s built-in algorithms current, regardless of changes in the worldwide economy. As more information it is given, the more it learns and the more accurate it becomes with the result. Therefore, it recognizes patterns and will give updates as relevant data is fed through. Deep learning is a subset of machine learning, which helps the algorithm learn without a human programmer. Through deep learning, information filters through a hierarchy that starts simple and gradually becomes more complex and specific. As the data is going through each level, the algorithm is able to determine what the object or data point belongs in what category. The process behind deep learning is essential to the success of machine learning. Since there is a grand amount of information that the algorithm must analyze to come to a conclusion, a human programmer would not be able to sit there and go through every single point – that would be extremely expensive and time consuming. Many industries and companies are beginning to use machine learning, to cut costs and become more efficient. Some of those industries include transportation, advertisement, medical, education and entertainment.

Transportation & Advertising – Antoneta Sevo

One industry that is being hugely affected by machine learning is transportation, particularly by autonomous vehicles. Though taxi companies like Waymo and Uber are implementing this technology within their services, companies like Tesla are creating vehicles that will be available to consumers for purchase. Tesla has introduced a system called Enhanced Autopilot that allows the driver to sit in front of the wheel and do nothing while the vehicle operates itself. The system includes features like active cruise control, forward collision warning, match speed based on traffic conditions, change lanes without the driver’s input, merge on and off highways and park itself. One main goal of Tesla is to successfully have a car drive itself across the country from LA to New York. Machine learning works by interpreting the ever-changing scenery detected by the sensors surrounding the vehicle. Once the technology is perfected within the industry, there may be no limitations to the future of driving.

As machine learning continues to be introduced into different industries, the amount of information being collected about users and consumers continues to increase. These algorithms are created to handle and sift through large amounts of data in order to identify a pattern and produce a result based on the patterns. This is beneficial to companies for advertising purposes so they can recognize habits of consumers. Essentially, companies track your browsing history to suggest products or other websites. They are also able to figure out how much the user can afford and when they might be able to afford it based on their demographics and when they usually purchase products. Since it would be time-consuming and expensive for a human to find those patterns, an algorithm is put in place and works in real-time. It is a much more efficient way to produce targeted advertisements. The more a user searches and buys, the more information the program collects, and the more accurate the advertisements become. By using machine learning, Target knew a teenager was pregnant before her father did. The algorithm noticed a pattern in her internet searches and sent advertisements designed for pregnant women to her house. That was how her father found out. Machine learning can collect an endless amount of data in order to have accurate predictions, and in some cases, they can be too accurate.

With new technology comes ethical issues and implications. When it comes to autonomous vehicles, many things can go right and there would be a smooth ride, but many things could go wrong as well. For example, the algorithm may make the wrong decision, misinterpret an object, or can be hacked, causing an accident. In order to avoid the program from making the wrong decision or misinterpreting, quantum computing must be implemented so the algorithm could work faster. It is important for companies to consider full transparency, so people understand how the car operates and how it is protected. This also means certain types of cybersecurity and encryptions should be put in place to protect the car from being hacked and causing harm to people on the road. Most car companies do not operate in an ethical manner now, however, they should begin to think differently in the future. There are also many different ethical issues that arise through the advertising industry. Once people find out they are being tracked and targeted, they begin to feel uneasy, as they should. Companies should allow consumers to opt out of their services so users will feel more in control of their information. Also, it would be beneficial if consumers were more aware in order to identify what these big corporations are tracking so they can do things differently to take themselves out of being tracked. Consumers can use a Virtual Private Network, use a search engine that does not track them like DuckDuckGo, or they can choose to use an opt out service. There are plenty of options, the first step, however, is to be aware that something is wrong.

The final scare that comes along with new technologies in machine learning and Artificial Intelligence is unemployment. Many articles talk about how many blue-collar jobs will be obsolete in the coming years and most of them are not wrong. It is a topic that those who want to be employed have to know about. Some jobs that may be affected in the future are schoolteachers, taxi or commercial drivers, positions in the medical field, etc. This technology has the potential to affect an abundant amount of jobs across plenty of industries, which means a solution must be in place before mass unemployment occurs. A short-term solution may include technology companies implementing training programs for current employees to learn the new (and necessary) skills. Another solution might be considering a universal basic income for those who are unable to find work or for everyone if machines and Artificial Intelligence take a significant amount of jobs. In time, as more and more jobs disappear, we, as a society, will have to redefine the definition of work. Millions of jobs will disappear and we will have to accept the change. In order to survive the inevitable societal and economic change, we need to accept that it is happening now. The next step is to begin learning in an exponential way instead of a linear way. The way our world functions now will certainly not be the way it functions in the future and we need to prepare to face it head on.

Medical & Education- Mayra Luna

Machine learning is all about data recollection and data processing. Its capability of analyzing huge amounts of data will eventually replace a great number of human jobs, even those that require a higher education. However, as explained by Medtronic CEO Omar Ishrak, the real value of artificial intelligence is in making more efficient use of human resources in healthcare.  For example, IBM’s Watson can read 40 million documents in 15 seconds, optimizing time and performance. Healthcare has loads of data: Test results, consultation notes, scans, appointment follow ups, etc., creating an ideal environment for the use of Machine Learning (More data, better results).

The ability of computer systems to assume tasks for humans has improved efficiency in healthcare. Now hospitals are getting into the game, deploying AI to take on challenges from diagnosing patients more quickly in the emergency room. Other technology developers, are focusing on software that can read CT scans and other medical images and then suggest the most likely diagnosis by reviewing similar images stored in patient databases. And these programs can accurately process these tasks far faster than human technicians. For example; Stanford researchers have developed an algorithm that offers diagnoses based off chest X-ray images. It can diagnose up to 14 types of medical conditions and is able to diagnose pneumonia better than expert radiologists working alone.

Machine learning can help healthcare executives and caregivers with things like precision medicine. Precision medicine is “an emerging approach for disease treatment and prevention that takes into account individual variability in genes, environment, and lifestyle for each person.” This approach will allow doctors and researchers to predict more accurately which treatment and prevention strategies for a particular disease will work in which groups of people. It is in contrast to a one-size-fits-all approach, in which disease treatment and prevention strategies are developed for the average person, with less consideration for the differences between individuals. For example, tech giant Microsoft wishes to improve health care using machine learning and AI. As part of this initiative, Microsoft is expanding into cancer research and treatment, and its approaching cancer cells as if they were a glitch in a computer system.  

Though ML promises to drastically improve the efficiency and effectiveness of healthcare, when it comes to predicting, diagnosing, and treating medical conditions there are many concerns: Data quality, Manipulation risk, Obscured logic

Four factors should be present to improve accuracy and overcoming risk: Confidence scores, Complex rules, Clinical data, Natural language processing

We must understand that machine learning is a powerful tool, not a complete solution. There is no substitute for a skilled physician’s expertise, however with the right data, ML can certainly help accelerate diagnosis, treatment, and program development.

Machine learning systems can give teachers more free time to actually teach and mentor on a more personalized level, instead of focusing on the never-ending grading and lesson planning. In addition, the current learning system consists of a “one size fits all” model. We are in desperate need of a system where the educational experience can be personalized to each students abilities and needs. Machine learning could address these issues by collecting and analyzing the data generated by students, identifying meaningful patterns and transforming that information into structural knowledge. In other words, when the student interacts with a digital learning platform, the machine learning system can accurately predict and better assess that person’s educational level, being able to tailor specific material for that individual.

Specific roles and examples of Machine Learning and its use in education: Content analytics, learning, analytics, dynamic scheduling platforms, grading systems, process intelligence, predictive analysis.

The ethical issues may include:

  • Unemployment: Labor is concerned primarily with automation. As we continue to implement machine automation, certain jobs (predictable physical work) could disappear. While this might sound like something bad, we must also keep in mind that new jobs will be created due to this disruption. The real question is how are we going to educate people for these new jobs?
  • Inequality: By using artificial intelligence, companies can drastically reduce the reliance on human workforce, which means people will work less hours (therefore make less income, broadening the labor gap).
  • Humanity: Humans are limited, while machines have unlimited resources. How will machine learning affect the way we behave and interact with others? Will we soon be interacting with machines as if they were humans? Will machines adapt to cultural norms?
  • Algorithmic Bias: Blind spots or biases in the algorithms could lead to discrimination against certain types of people (As seen in AI systems that can tell if you are homosexual).

Entertainment- Olivia Finan

The role that machine learning plays in entertainment is very interesting.  A common belief among the class is that entertainment is an industry that will more or less remain untouched in the coming years by the disruptive technology. However, research shows that this may not be the case. We have already begun to see drastic changes in the industry throughout the years, specifically in music and cinema. AI systems are being created and coded to be able to sift through video footage and put together a 10 minute video using several clips in under and hour.  This is impressive timing when you consider how long it takes to watch hours of footage, edit, tag, and compile all those clips into one succinct video my hand.

Filmmaker Oscar Sharp and technologist Ross Goodwin fed a machine learning algorithm with a bunch of Sci-Fi movie scripts to see what new script it would spit out.

The trends that we think will see in entertainment are not necessarily what is coming. How this content is created is what will change how the industry progresses through the coming years and how it is affected by disruptive technology. The entertainment industry is a prime example of how we must pivot and change how we do things rather than change what we do.

More so then ethical issues, legal issues come into play when talking about the entertainment industry. Laws regarding copyright infringement are always looked at thoroughly in this particular industry. However, ethical issues tend to revolve around the consumer rather then the company or a performer. For example, at live events, issues about safety come into play. This means that a major role would be ensuring that concert or festival goers are safe at the venue. This is just one example of what time of ethical dilemmas companies have to deal with in the entertainment industry.  What these companies now have to begin thinking that the ethical implications that once were may not be the same anymore. We may not be having live concerts, there might be other ethical issues that arise, and that is what the industry predicts.

The entertainment industry is not exempt from responsibilities associated to employment. However, if there is a change in music, for example, and humans are no longer the ones producing it then what happens to the people who world at record labels, or what happens to the artist themselves? This section of the machine learning project will explore all the changes the industry sees coming in the future and how the major players are beginning to prepare and adjust to the changes.

There Is No Timeline For Machine Learning

Machine learning has the capability to transform our future and then some. Currently, humans are working towards building machines that learn from themselves which could ultimately become so good at teaching themselves that it could eventually eliminate the human aspect. In today’s society, machine learning is becoming more and more useful across many different industries. We know what is happening now, and what could possibly happen in the nearest future, but what about the future in general? This answer is a little foggy. It is extremely difficult to foresee what machine learning and artificial intelligence will be able to accomplish. There is an endless amount of opportunities and it is a technology that is not fully understood just yet. This means the implications and the endgame for machine learning and artificial intelligence is unclear.

In order to prepare for the emergence of this developed technology, it is important to imagine the craziest possibilities and be ready to implement what is necessary to adapt our society. One article, written by Cade Metz, talks about the possibility of artificial intelligence being able to create artificial intelligence on its own. This idea stems from the reality that only about 10,000 people have the skills and knowledge needed in order to produce such innovative technologies. Google is behind AutoML, which is already successful at building an algorithm that can identify objects in photos more accurately than programs built by human experts. Eventually, computers may be able to create Artificial Intelligence by using a more advanced machine learning. It is already happening, and it is only a matter of time before human experts and Artificial Intelligence work side by side on a daily basis.

Today in class, we had the pleasure of witnessing a presentation by Frank Diana, and one point he made really stuck out to me. He mentioned how many people question when they should start preparing for the era of machine learning and Artificial Intelligence. The simple answer is now. The more complicated answer is that there is no official time to start preparing. This technology has already implemented itself into our lives without some of our knowledge. It is there when we are shopping, use a GPS app, use an entertainment service, and many other places. The thing is many people are infused with the convenience, therefore, they do not notice why or how it is happening. This means that those people are not prepared for the future of Artificial Intelligence and machine learning. Those who are educated in the subject are aware at how much guessing is going on. No one truly knows what is capable of happening or when it will happen. Most people will ask for a definitive timeline for the future, but the truth is, no one knows. This is either amazing, terrifying, or both. That is typically the way these revolutions happen. There are those who expect and accept change and there are those who simply ignore it. I found it amazing that machines and algorithms were able to learn from past mistakes but now there is Artificial Intelligence building those algorithms. It is simple mind blowing.

The main thing to take away from learning about machine learning is that development is nowhere near finished. There is still plenty to be discovered and the guessing game on time will continue. The key to the future is acceptance, the ability to adapt and a crazy imagination. This new revolution is happening and those who are aware will be better off than those who ignore it. Innovations, like Artificial Intelligence building its own algorithms, are popping up every single day and it will not slow down anytime soon.

Machines “Learning” How To Steal Our Jobs

One of the biggest concerns many people have are whether their jobs will be taken over by the new and improved era of programs and robots. For the past few decades, plenty of jobs have become automated and this is what people who need to put food on the table are most afraid of. Some jobs that have already been affected, which is making the switch a little more real. Those may include or will include jobs in transportation and logistics, office support staff, sales and customer support staff and many more. These specific types of jobs will be replaced by autonomous vehicles, computerized check-ins for buildings, and chatbots and machines. Each of these new technologies have the ability to change the whole concept of unemployment. The question is, in what way will society and jobs as we know it have to adjust?

There are a few other career paths that may also be in jeopardy that some never thought of. For example, in the TedTalk below, Anthony Goldboom talks of a challenge presented by Kaggle to build an algorithm that could successfully grade high school essays. The winning team built one that was able to match the grade that a human teacher would give. This means that physical teachers or institutions may not be needed. As a society, we are already seeing college classes and some colleges being digitized and online. A machine is and will be capable of doing more work at a faster pace than humans do. This will only get more advanced as time goes on and as the technology progresses. Machine learning has the capability to adapt and become skilled in many different industries, which suggest that many jobs will be lost but overall efficiency can increase. Another example lies in the medical field. Many systems are able to spot diseases and tumors 50% more of the time than a human can. This means that these machines have a higher chance to potentially save more lives than current doctors or surgeons. The precision and certainty of this technology reduces human error, which can only help society.

Though many jobs will be replaced or assisted with machines and Artificial Intelligence, there will be those that require programming the technology or anything related to that matter will still be needed. The question then becomes, how do you prepare the current or future unemployed and employed to successfully adapt to the new world of jobs? There are potentially two answers. One is that current technology companies can implement a training program to help people learn new sets of skills. This means that we, as workers, would be redefining what it means to work. That is what people have been doing ever since automation started decades ago. When automation seemed to be taking over, new jobs presented themselves and more opportunities. By preparing, the working class now for a new future would be beneficial because this new era of technology will redefine our lives as we know it. We all will be impacted in one way or another and it is best to be aware of the change that is coming. The key may be to not only change ourselves but also redesign the system. This brings the second answer to mass unemployment, if inevitable, to light. Another solution could possibly be a universal income in order to help those who do not work to become completely lost. If plenty of people are out of work, with no money coming in, how will they live? What if there are no more jobs to adapt to? If a universal income were implemented then every citizen would have some sort of income and rise above the poverty line without a job. This also means the man who lost his job as a taxi driver would still be able to provide his family of five. Additionally, it could provide people with the opportunity of a passion filled life without stress of a job. It would be quite a long process to redesign the entire system of income for an entire nation, however, with the rise of automation and Artificial Intelligence, the government may not have a choice. This decision may have to occur in the next few weeks, months or years, but it would be valuable to have an idea of a solution before it is too late.

A TedTalk given by Garry Kasaparov, addresses how humans should work with intelligent machines instead of shying away from and ignoring them. A machine, IBM’s Deep Blue, was able to beat Kasaparov, who was a world champion, at chess in 1997, its early stages. He claims that what could be accomplished by computers or humans alone could be even more successful if put together. Imagine how precise your next doctors appointment could be in the near future? By normalizing the training of employees to program and use this new technology efficiently, means creating a world that operates at a new level we have never seen before. The modern era of machines and Artificial Intelligence indicates that our whole system of employment and income must adapt. We must modify and accept the upcoming society of work and redefine what it means to operate side by side with intelligent machines.

The impact of Quantum Computers on Society

 

Similarly to our conversation on artificial intelligence, it’s safe to say that the first to fully create controllable quantum computers will be thousands of years ahead of his counterpart. One of the problems is that the very people that we do not want to have this type of power are the only ones who can afford the technology right now. Imagine the power that Google will have access to with the exponential growth that they are experiencing in the field of quantum computing. Match this with their advanced technology in the field of artificial intelligence and you have a match made in heaven. As our Machine Learning group has mentioned: we will have to face the fact that there will be a considerable number of jobs that will be replaced by artificial intelligence. To add to the issue, with the advance of quantum computing technology, this is a problem that we will have to address much earlier than we expected. If the only thing machines require to complete any given task is time to learn the task you have given it, which could be accelerated with the computational speed of a quantum computer, then how long will it take before there are machines working in hospitals operating with little to zero percentage error? Or even imagine if the power and speed of quantum computing was used to advance the field of gene therapy how fast could we engineer the perfect human that never dies?

The beneficial implications and uses of this technology are innumerable, but anything of such reward comes with a high level of risk. Longer life expectancy leads to overpopulation, which is a problem that we are not ready to face. We have also never faced any being of superior intelligence to us, which will be an issue of its own. We must make sure that this technology is placed in the right hands and is used for the beneficial advance of humanity. This has to be done appropriately the first time around because any slight error could threaten human existence as we know it.

Speaking of, Temporal Defense Systems recently purchased one of the most powerful quantum machines available and they will help build the basis for the security systems that will be necessary when this technology arrives on a larger scale in the near future. With quantum technology, they will be able to take security and defense to the next level with device to device authentication, increased speed of threat detection and real time security level rating. James Burrell, exclaims that there are considerable benefits associated with this technology, which are mainly their ability to solve more complex computational problems. This allows them to increase security on pace with the ever changing operational network.

As you may recall from previous posts, quantum technology works on an exponential scale. Because the basis of this technology is currently being made a reality and put into practice, progress will only rise exponentially from here. Remember all of the issues that we brushed off because they were too far ahead of their time? Well the reality is that these problems will be in our face before we know it and we must be prepared to take them on with viable solutions once they are here.

https://www.wired.com/video/what-is-quantum-computing

Machine Learning In Our Everyday Life

Related image

When we think about artificial intelligence we might imagine something out of a science fiction movie. However, we might not realize that in one way or another we are using some type of machine learning in our day to day routine. Machine learning is a form of artificial intelligence which allows computers to learn from examples rather than having to follow step-by-step instructions. We encounter machine learning systems daily through our smartphones and our computers, and whether we realize it or not we have become dependent. One thing that’s for sure is that machine learning is already part of our everyday life.

The following are just a few examples on how we constantly interact with ML:

  1. Commuting: Through something so simple as using our GPS or taking advantage of online transportation services such as Uber or Lyft. Using location data from our smartphone, Google Maps can analyze the speed of movement or traffic at any given time. While we are driving, our current locations and velocity are being stored in a central server for traffic management. Our data is then used to build a map of the current traffic, thus giving us route suggestions and estimated time of arrival. As for App based transportation services such as Uber, machine learning systems are used to determine an estimate price of your ride, to minimally optimize your wait time, and to optimally match you to the best service.
  2. Virtual personal assistants/Voice to text: A standard features of smartphones today is voice-to-text. Using spoken commands to ask your phone to carry out a search, or make a call, relies on technology supported by machine learning. Virtual personal assistants (VPA) such as Siri, Alexa, or Google Assistant are able to follow instructions due to voice recognition. Machine learning is an important part of VPA since they collect and refine information, and later that data is used to render better results in accordance to our preferences.
  3. Social Media: From personalizing our news feed to ad targeting, social media platforms use machine learning in many ways. Facebook uses AI for facial recognition, so when you upload a photo, faces are automatically highlighted with suggested friends to tag. Instagram, which FB acquired in 2012 uses machine learning to contextual the meaning of emojis. As for Snapchat, the facial features (lenses) track facial movement allowing us to use animated effects or masks that adjust to the movement of our faces
  4. Online Shopping: Think about how Amazon suggests certain products as “customers who viewed this item also viewed”, machine learning is the technology that helps deliver these suggestions through recommender systems. By analyzing data about what customers have bought before, these systems can pick up on patterns in purchasing history (On the basis of our behavior, items liked, items added to our cart, etc).
  5. Music & TV Streaming Services: Recommender systems are also used to suggest movies or TV shows on streaming services such as Netflix. These systems use machine learning to analyze viewing habits and on how we rate that show/film. Music streaming services (Spotify/Pandora Radio) also use machine learning to suggest music.

 

The Incredible Journey of Machine Learning

Machine learning has a great amount of meanings to different people. It has been around for years and now it is getting the attention it deserves. The technical definition of machine learning is “the concept that a computer program can learn and adapt to new data without human interference. Machine learning is a field of artificial intelligence that keeps a computer’s built-in algorithms current regardless of changes in the worldwide economy.” Many people think of artificial intelligence when they hear machine learning, and those people are not wrong but they are not right either. Essentially machine learning entails programming but instead of using code, one uses images and videos to build a model. A training set is selected, and the programmer chooses what imitates a positive or a negative for that specific set. Then they assign a specific model to use. The data one works with effects the model type and the size and quality of the sets. Additionally, it impacts the outcome. The process and technology itself is extremely complex and if something does not work correctly, the operation begins all over again. On the other hand, Artificial Intelligence depends on feedback as it interacts with the world and continuously adapts to new changes. There are layers to the overall aspect of Artificial Intelligence, which can easily be explained by this graphic:

Each level is a subset of the one before it, meaning one cannot properly function without the other. Deep learning carries out the machine learning process using an artificial neural net that is composed of levels arranged in a hierarchy. The particular network will learn something simple at the primary level, and then it sends the data to the following level. At the next level, the simple information is mixed with more data and becomes a little more complex. Then, it is sent to the third level. This process repeats itself at every level as the hierarchy becomes more and more complicated. Deep learning is essential for machines to be able to learn without a human labeling every single type of information that comes in, which is supervised learning. For instance, an endless amount of personal data is collected from social media accounts, hardware and software service agreements, app permissions and cookies. Businesses use this information for many reasons and it can be extremely valued. Not every element of the data is labeled which means it cannot be used to teach the machine learning programs that depend on supervised learning. To have a programmer oversee every data set that comes through would be extremely time consuming and expensive. Luckily, deep learning can help. It excels at unsupervised learning where the data is not labeled. Overtime, with enough information, deep learning can determine what is what, which helps machines learn without a human programmer.

Shockingly enough a children’s animation movie illustrated the abilities a robot could possess through machine learning. The Incredibles, a Pixar movie introduced in 2004, provides a good visualization of how machine learning and artificial intelligence work and the speed at which it processes real time information. Over time, it learns how Mr. Incredible fights and is able to determine ways to beat him. The villain, Syndrome, designed these killer robots, The Omnidroid, to eventually eliminate Mr. Incredible. However, over the years he put his prototypes up against other Supers in order to perfect his creation. As Mr. Incredible goes up against The Omnidroid the first time, it learned how he moved and predicted what he would do next. By doing so, it was able to delay Mr. Incredible’s victory. Even in 2004, moviemakers implemented this advanced technology in a children’s movie in order to show its future capabilities. A machine being able to learn in real time. As you watch the video below, pay attention to how Mr. Incredible initially jumps over The Omnidroid and how it then predicts him to jump over again.

Google has successfully combined deep and machine learning into a new program called AlphaGo. Go is an ancient Chinese board game which is allegedly the most complex and subtle of all board games. This is not as typical as chess, where a computer can calculate every possible move thousands times fast than a human. In Go, the possible moves are endless, which means no computer can handle trying. However, researchers in the UK were able to teach a computer program to play Go, but it also has the ability to learn from its mistakes. It used deep and machine learn to study which moves left it vulnerable and which ones lead to victory. In the beginning, there were a lot of mistakes made, but this was a good thing because the program was learning. The more it played against humans, the more it learned. As time went on, it played against more skilled players, and in addition with those games, it played matches against itself. When the researchers put AlphaGo up against itself, the program became much better as it was able to compete against a player at the same skill level. Eventually they put AlphaGo against the number one player in the world, Ke Jie, and beat him in five games out of five. After that, there was no one else left to beat in the world. AlphaGo was able to learn from itself and become the greatest Go player in the world, and it is just a computer program. If it is already the best, how much better could it get? AlphaGo Zero.

The opportunities to use machine learning are endless and highly effective. If in 2004 a children’s movie can display its basic capabilities then it is astounding to see how far it has come since then in the real world. The news of AlphaGo and AlphaGo Zero are surprising because it was not expected that this technology would advance this rapidly. These machine learning programs is truly a breakthrough breakthrough and it will continue to improve in the future.

Evolution of Machine Learning

The history of machine learning begins in 1950 with Alan Turing and the “Turing Test”, a test he created to determine whether or not a computer has real intelligence. In order for a computer to pass the test, it must be able to fool a human into believing it is also human. The test basically was a guessing game. There was a judge, a female contestant and a male contestant and the based off the answers to a series of questions the judge had to choose which of the respondents was human and which was the computer. The most recent modification to the test however doesn’t involve contestants. Simply, it is a judge and a computer and the judge has to determine if the respondent is a computer or a human. Since Turing introduced this test, the results have proven to be highly influential and also regularly criticized. However, overall the test has become incredibly influential in the artificial intelligence and machine learning world.

Another major advancement that occurred in the machine learning world was in the 1990s when scientists began creating programs for computers to analyze. From the data, computers where able to draw conclusions based on the massive amounts of data they were presented with. Simply, the computers were able to “learn” from the data. Computers, like IBM Watson are now able to store endless amounts of data, ranging from different topics, and they are able to make connections between the data in order to draw conclusions. For example, in 2011 IBM Watson was put to the test. Watson went up against the two greatest Jeopardy champions. After the first round, Watson was ties for first with $5,000. However, Watson made an incredible comeback and entered into final Jeopardy with $36,681, the next player having only $5,400. This test was groundbreaking not only for the IBM team, but also for AI and machine learning. Watson was able to enter into its data base to collect data it needed to answer the question, analyze the data, and great an answer all within seconds.

In 2012, Google developed an algorithm that was able to autonomously look through YouTube videos to identify all the videos that contained cats in them. Similarly, in 2014, Facebook developed the DeepFace software which was an algorithm that was able to recognize and verify individuals based on their photos.

These singular advancements in machine learning have made huge impacts on technology today and how we use it daily. Seeing how computer intelligence went from the Turing machine up to iPhones with facial recognition is incredible and it just shows that it will continue to change rapidly and that there is still much more to learn.

Ethical Challenges in Machine Learning: A look at Advertising and Ethics

Image result for machine learning and ethics

There is no doubt that Machine Learning systems are transforming our lives, creating both a positive and negative impact. Companies such as Amazon, Facebook, IBM and Microsoft have taken advantage of the boundless opportunities artificial intelligence has to offer. However, businesses and governments are relying more and more on ML systems to make important decisions, raising questions about fairness, ethics and morality.

Despite the “optimism” that surrounds AI, we must be aware of its potential outcomes and the ethical challenges. Increasingly, technologists and policy makers are dealing with these questions, asking themselves how can AI and individuals ethically interact.

Some of the issues we may face include:

  • Unemployment: Labor is concerned primarily with automation. As we continue to implement machine automation, certain jobs (predictable physical work) could disappear. While this might sound like something bad, we must also keep in mind that new jobs will be created due to this disruption. The real question is how are we going to educate people for these new jobs?
  • Inequality: By using artificial intelligence, companies can drastically reduce the reliance on human workforce, which means people will work less hours (therefore make less income, broadening the labor gap).
  • Humanity: Humans are limited, while machines have unlimited resources. How will machine learning affect the way we behave and interact with others? Will we soon be interacting with machines as if they were humans? Will machines adapt to cultural norms?
  • Algorithmic Bias: Blind spots or biases in the algorithms could lead to discrimination against certain types of people (As seen in AI systems that can tell if you are homosexual).

So what precautions are being taken to stay on top of these issues? Recently DeepMind created a special research unit devoted to investigating ethical issues surrounding Artificial Intelligence. The people at DeepMind believe that these systems should remain under human control and be used for socially beneficial purposes. Elon Musk launched the “OpenAI” Institute, which is a non-profit research organization that “aims to promote and develop friendly AI” in such ways that will benefit humanity as a whole.

Also, a partnership on Artificial Intelligence has been formed by Amazon, Facebook, IBM, Google, and Microsoft (Apple joined at the beginning of this year). As best described in their web page, this partnership was established to advance the public’s understanding of AI, as well as formulate the best ways to implement this technology.

Image result for partnership on ai

 

Ethical Impact on Advertising

Artificial intelligence isn’t inherently good or bad, it is in the way we use it that determines its nature. Machine learning algorithms come into use across multiple fields, such as advertising. AI systems have the ability to collect and analyze data to determine human behavior. Every time you click on a link, google search something, every like you make on Instagram or Facebook, your location data from your cell phone, even your credit card transactions can give huge insights into a persons habits.

The advertising industry is obsessed with understanding human behavior, therefor taking advantage of these machine learning systems. However, as discussed in this article, AI raises many questions in terms of ethics. Imagine advertisers knowing us better than we know ourselves? The more they can understand us as individuals, the easier it will be to persuade us.

http://http://www.slate.com/articles/technology/future_tense/2012/01/behaviorally_targeted_ads_and_the_ethical_dilemmas_behind_building_consumers_into_ads_.html

The article above mentions the ethical implications of adding consumers into commercial advertising. In other words, through facial recognition and other technologies, behavioral marketers will be able to access your Facebook pictures, create a 3d image of you and insert it into ads. This obviously raises various questions on privacy, consent and data security. Imagine walking by a billboard with an imagine of a person on the beach. Now imagine that the person on the billboard consequently takes your physical characteristics. Creepy? Yes!! However, as with every other new disruptive technology, consumers will eventually learn to adapt to these marketing techniques.

 

 

Ethical Implications of Autonomous Vehicles

Autonomous vehicles will one day become the normal way to travel. However, there are still plenty of ethical issues and dangers this new technology holds. It is easy to forget about the impacts technology has on society because everything is innovative and exciting. Nevertheless, the ethical implications must be discussed in order to avoid potential accidents or tragedies. The possible dangers might include the likelihood of a hacking or if the vehicle does not make a decision fast enough. Some of the ethical effects can be what kind of decisions should be made in certain situations and if programmers should have transparency with consumers, so they know exactly what the autonomous vehicle entails.

These cars rely on machine learning in order to evaluate situations and make a decision. The computer is simply not given a set of rules to follow. Instead, it is fed images of objects – for example, a pedestrian, a ball, another vehicle, etc. – and tries to guess what that object is. In the beginning, it will guess wrong. However, as time goes on, the program adjusts itself, and continues to try as more information it is given, it begins to learn what is what with the help from the cars sensors. If there is unidentified object in the road, the car should reduce speed or stop altogether. Another way for the vehicle to learn is by feeding it certain traffic information and the programmer would say the right way to get out of it. The algorithm will learn from that, along with other aspects of a situation, and determine the correct way to get out of it. One of the main issues is how fast these computers will be able to make decisions. While humans drive, it takes a split second to make a mistake and terrible things could happen or it takes just as long to avoid that. Most people have good reflexes and are able to avoid a tragedy. Autonomous vehicles must have that same ability to avoid accidents. Machine learning and quantum computing must work together in order to allow those cars to make effective and quick decisions. The programmer must teach the computer the basic rules of the road and allow the machine to learn impactful and successful ways to avoid accidents.

Another issue that stems from autonomous vehicles is the possibility of someone hacking the program of one. Skilled hackers are able to break into almost anything, so how is a vehicle any different? Software and algorithms will power them, which can be susceptible to malicious people who have the intent to harm others. It might not be a trend now, however, as this technology becomes more normalized, anything is possible. Car and technology companies claim that the consumers will be safe from hacking, however, it would be in the best interest of the consumer to not take their word. Some in the industry have waited to announce a recall until it became cheaper than paying the wrongful death lawsuits. It would take enough steps in the right direction for car and technology companies to act in an ethical manner. It would be best to not trust their word. As the conversation about hacking becomes more popular, there should be laws requiring cars to have certain types of encryption and cybersecurity in order to protect the passengers of autonomous vehicles. These companies should also consider the approach of a greater transparency with their consumers. That way, if they are ethical and genuine, consumers could trust using their products, in this case an autonomous car.

Though autonomous vehicles will be able to solve many problems and subtract human error, it is important that it does not replace it with programming errors. It is also essential that the algorithm learns in an ethical and correct way. It may take time to determine what that is, however, the minimum amount of accidents and casualties is desired. Car and technology companies must implement sophisticated cybersecurity to protect against any type of hacking that might take place for any reason. There is a long road ahead when it comes to creating the software in the right way. That way when it learns, it learns in a correct and moral method. Machine learning plays a big role in the success of autonomous vehicles and though there will be problems, it would be beneficial to minimize them as much as possible.

Machine learning debut in the entertainment industry crashes and burns?

I work at a market research firm and ever since the semester started I’ve begun to think about how disruptive technology will affect the particular industry I work in currently. Within my firm I work in the media and entertainment department which is why when it came to choosing what industry I wanted to focus my machine learning project on, I jumped at the opportunity to focus on entertainment. Spencer made a good point in one of our first classes, that when machines take over and people are out of jobs, there will always need to be a solid entertainment industry to keep people busy.

Google gives us a goo idea of what entertainment is like for us now because of the machine learning that how technology uses. It creates a more personalized experience for us and allows us to watch our content when we want.

According to Forbes, there are six major digital transformations in the media and entertainment industry. Multi-channel experiences are the norm now; Accenture Digital did a study that showed most people obviously use different devices to watch certain things, but often, people are viewing content on their devices simultaneously. Creators are becoming scared that it’s not necessarily the content that people care for, it’s more about the convenience.  This is where the AI comes into play. AI is getting more and more creative, according to Forbes. The technology is used in the entertainment world in many ways, one being to create plots of shows and movies based off of box office ratings. This new wave of computer-human collaboration is already working effectively within the industry. The result of this collaboration and integration of AI in the entertainment industry is that the computer will learn how to collect box office ratings and Nielson TV ratings data on its own to then create a plot and compile a final trailer for review all within 24 hours, which is significantly less than the average 30 days spent editing manually.

A prime example of machine learning specifically in the entertainment industry is located within Netflix. Based on what you watch, Netflix recommends shows or movies that are similar to the shows you have watched or are currently viewing. To that point, in 2013, Netflix released “Max” your personal recommender on the app. However, this feature crashed and burned due to the fact that “Max” didn’t sync properly with consumers Netflix accounts making is results less and less accurate and his recommendations poor. This sent Netflix back to the drawing board to see how to use machine learning to their advantage.

Similarly to Netflix, most companies are researching how to benefit from AI and machine learning as both technologies become more prominent. There is a lot of research and development regarding machine learning for entertainment companies and they seem to have their finger on the pulse, as of right now.