Category Archives: Week 12

A Call for Regulation or Education? Reframing AI Discourse

The leading issues within AI discourse stem from a lack of transparency and information surrounding emerging technologies. In tackling some of the core categories within AI, it is useful to start by looking at media representation, which is where the general public go to get their knowledge on a given topic. 

From here, we can ask: what information do we actually have concerning a topic (Deep Neural Networks, Natural Language Processing, etc)? Where is the information coming from and who is the intended audience? Is it an article that is meant to advertise a product rather than to detail out accurate information?

During this course, we have covered key concepts that help de-mystify AI systems and also lead to more informed questions that can help us move forward in both our analysis and our understanding of the current climate surrounding these topics.

Reframing AI Discourse

One of the important distinctions to be made is that although AI is meant to “do the sorts of things that minds can do,”  the intelligence (or program) does not have an ability to think for itself (Boden, 2016, pg. 1).

Data and Representation

  • Everything depends on computable data, the context requires that it be defined as a certain data type
  • Deep Learning requires large amounts of data, where a human can recognize patterns more easily and accurately due to context
    • Deep learning is not deep knowledge or accumulated knowledge, but rather deep layers used within the network

Autonomy of Computational Artefacts vs. Autonomy in Humans 

  • AI systems are not autonomous. Machines are not like humans in that they do not have their own interests and freewill to act as they choose, debunks the myth that machines or robots will decide to take over human lif
  • The more important question to focus on: Who is creating AI systems (especially autonomous weapons + other tools for destruction) and has control over these artefacts?

Limitations of Neural Networks / Machine Learning 

  • Although Deep Neural Networks can be very useful in picking up pattern in large amounts of data, one of the main issues is that they don’t tell the whole picture. They can only select data based off of set parameters and rules – which does not translate into human problem solving or decision making.
  • A ConvNet is a large collection of filters that are applied on top of each other, they will learn to recognize the labels that we give them
  • Context is important when comparing what a neural network can do vs. a human brain
  • Algorithms are designed by humans, therefore a computer or AI system is not inherently biased. This is a big theme because a lot of media exists on discussing how AI is evil or bad — which takes the attention off of the systems we have in place and the algorithms we design. At the same time, when it comes to regulation — a big issue is that we don’t know all of the steps involved in the process of a deep neural network making a classification. This is due to the complexity of the system and hidden layers. 
  • Machine Learning is only one method of learning within AI Systems, and therefore should not be the only focus when looking at ethical implications of AI systems

The EU’s Set of Guidelines for Trustworthy AI tackles many of the ethical issues we have discussed throughout this course: privacy and data governance, transparency, human agency, bias, environmental well-being, and accountability. At the same time, these guidelines expose the problems implementing the regulation of AI. In learning about the ways AI is designed and programmed, it’s clear that some of these regulations are still too vague to be effective or implemented without becoming censorship. They are also very general guidelines which continues the cycle of blanket terminology and generalizations used by tech companies to “explain” their products, services, and policies.

Given this current example, we can see there needs to be more discourse surrounding the ethics of AI that is open and inclusive to creators and users alike. The ethical implications continually point to a need to understand what we are creating and the consequences of those creations to society as a whole. Software engineers and developers are not the only ones who need to be concerned with learning how emerging technology works — additionally, the method of teaching engineers only the mechanics and technical aspects of the process is in itself an ethical issue. Education, even more than regulation, will be necessary in order to move forward in creating systems that are safe and ethical. The more the information remains in the hands of a few corporations, the more likely we are to have a digital divide, and without the resources to teach / inform people about how AI systems work (due to IP issues etc) + how to develop alternative systems – we are stuck making guidelines without the means to implement them.

 

Machine Translation & Data Privacy

Tianyi Zhao

Artificial intelligence, although in the fast-growing stage nowadays, is still a blackbox waiting for exploring and exploiting. Currently we are on the stage of leveraging with neural network, in which machines can learn advanced algorithm from practice and testing. The key fields in AI that have mostly impressed me during the course are machine learning and natural language process. The typical practice that combines these two is machine translation. As deep learning develops, neural network has been applied to machine learning and replacing the previous statistical one. With the encoder-decoder model, the source sentence is encoded into a fix-length vector from which a decoder generates a translation during the translation process. It associates context to find more accurate words and automatically adjusts to a more natural sentences syntactically that are smoother and more readable. Google Translate realized its transformation from statistical machine translation to neural one with multiple input methods in 2016. However, the technology still has problems in sequence and wording in reality. Besides, pattern recognition has also been applied to machine translation. There are mainly two types—image and speech. The multimedia in the source input is acceptable, however the output is always in text. Personally, I think the next step of machine translation is not only the accuracy improvement but also the diversity of output. In the near future, there may not simultaneous interpreters any longer.

Besides, to improve the accuracy of machine processing outcome, there needs to be Big Data applied. So here comes a prevalent issue of privacy. How can we guarantee the data practiced for machines are collected legally or authorized? There has been numerous data abuse scandals in the tech giants. During a research on Google Translation URLs, a police investigator was discovered to translate requests for assistance made to foreign police forces. The confidential information becomes no more “confidential” because of online translation. Don DePalma of Common Sense Advisory warned that “free machine translation tools such as Google Translate can in advertently result in a data leak.” (Brown, 2017) As machine learning becomes more popular, how can the public users do to protect their data when enjoying the comfort and convenience brought by machine learning?

Works Cited

Brown, Claire. “GDPR: Beware Data Leaks via Online Search and Translation Tools.” Today Translations, Oct. 22, 2017.

Big Data and Buyers Remorse

The negative impacts of big data are obvious especially as it relates to privacy and discrimination.  The more feedback that users give to social media and the corporations that control the internet, the more information that is at risk. Similarly, the more information that is available about different demographics, the more evidence that corporations and governments have to mistreat or treat people differently.  Big data is gathered through the inputs and feedbacks that are integral to operations within the digital world. Engagement with websites and digital media is a relatively seamless data gathering technique that benefits corporations and users ( though the scale of benefits is not even).

Many are well aware of the constant tracking of data, especially as it relates to our queries and behaviors on the internet. This tracking is often manifested in targeted advertisements that seem to follow users from website to app and any other digital place. Users willing give data and feedback to systems on a daily basis, but how does this feedback impact buyers remorse. People regularly purchase clothing and other times with few or very little thought or knowledge about the buyer or item. Though cookies and advertisements are annoying to users, when an item of interest is advertised to the right user, the user is empowered with more information about the item. Do tracking cookies and advertisements negate some aspect of the user’s buyer’s remorse?

Buyer’s remorse is the negative feeling of unmet expectations after a person purchases something. This feeling is often felt after online purchases which may appear as higher quality than the reality. Though pop up advertisements are not informational, advertising has shifted from pop-ups to more native advertising. These native ads are often produced and acted out by “influencers” who have gained acclaim through blogging, vlogging or instagram posts. In native advertisements produced by influencers, users are given informational reviews about everything from clothing to meal preparation boxes. Big data contributes to these influencers and native advertisements because it enables corporations to build these influencers. Influencers are trusted by their followers because of established ( no matter how true or not) relationship and reputation for honesty and thoughtful reviews(advertisements).

Native advertisements use just as much, if not more , big data as traditional advertisements, digital and otherwise. These native advertisements are almost always better received than  traditional ones, and are more useful to the viewer. Empowering users with more information about the brands and items they purchase from alleviates some of the negative impacts of purchasing items online, but further understanding of buyers remorse is needed to conclude the relationship between it and native advertisements.

 

It’s (Almost) All Hype

Over the course of the semester, the main theme that comes to the surface when discussing artificial intelligence is hype and the harms that this hype has on a broader conversation. No one really knows what artificial intelligence actually is, or what standard of intelligence to use to judge a machine’s achievement of such, and this confusion is reflected in media articles that work to obscure the sociotechnical systems that artificial intelligence systems are a part of.

Cycles of Hype and Fear — and calls for regulation on things that no one fully understands

Even before this course, I noticed that many different computational methods were conglomerated under the mantle of “artificial intelligence.” Any company that created anything began to implement apparent “artificial intelligence” in either its design or services. People responded to the hype train with excitement and capital, so much so that 40 percent of European startups that claimed to use artificial intelligence used no such methods. The claim that the organizations were somehow associated with artificial intelligence was enough to harness the hype beast into capital investment.

There are areas of genuinely exciting application for artificial intelligence – content moderation

While the massive hype train would allow an easy out for a cynic to dismiss all progress in artificial intelligence recently, that would be irresponsible as there are several ways and reasons that artificial intelligence is becoming an increasingly important societal conversation. Because there is so much digital content created and acted upon every day, hour, minute and second of the day online, there is a wealth of training data available to companies that create and train AI systems using techniques such as machine learning. Thus, with this recent accessibility of “big data,” artificial intelligence has improved markedly over the last decade. This improvement, especially with machine vision, has exciting – and ethically difficult – implications in the realms of automating content moderation online.

Sociotechnical blindness as a result of misunderstanding and hype

To me, the biggest takeaway in learning more about artificial intelligence and the coverage surrounding it is how deep and widespread the phenomenon of sociotechnical blindness is surrounding systems that utilize artificial intelligence. This concept, that Deborah G. Johnson and Mario Verdicchio introduced, explores the ways in which artificially intelligent systems are deemed to be separate entities of agency from their creators. Average people are unaware of the human-mediated design decisions that go into artificial intelligence and the systems on which these systems operate. That’s a symptom of – and why we get – simplistic headlines that say things like “AI is racist” or “AI caused a fatal accident.” This simplification of sociotechnical systems involving AI obfuscates human action and agency that goes into the system, making users feel powerless and allowing creators to eschew responsibility for real-world actions.

Deep Learning Challenge in Healthcare

Deep learning, as it is primarily used, is essentially a statistical technique for classifying patterns, based on sample data, using neural networks with multiple layers. (Deep Learning: A Critical Appraisal)

There is no doubt deep learning benefits our lives a lot. However, through this week’s reading, we know that deep learning and AI has lots of limitations. Take health care sector as an example.

According to Healthcare Tech Outlook, AI applications making waves in healthcare today, including disease detection at early stages, drug creation etc. The potential of artificial intelligence for making healthcare better seems to be indisputable, but the medical limitations of present-day AI has to be acknowledged.

First, there is the risk of feeding the computer with underlying bias. The outcomes of deep learning only depend on previous income data, so the forecasting and predictive abilities of smart algorithms are anchored in previous case. Therefore, deep learning might be useless in novel cases of drug side-effects or treatment resistance.

Second, this can be especially problematic since machine learning apps usually run as a “black box” where the machinations of its decision-making aren’t open to inspection. If a clinician can only judge a prediction based on a system’s final outcome, it may either undermine the human opinion or simply prove worthless.

Third, in problems where data are limited, deep learning often is not an ideal solution. (Deep Learning: A Critical Appraisal) Human beings can learn abstract relationships in a few trials, but deep learning thousands, millions or even billions of explicit training examples. Nowadays, patients’ digital data is not enough to implement deep learning.

AI works best in closed-end classification problems given that there is enough data available and the test set closely resembles the training set. However, our world is always changing and unstable.

Therefore, AI is not as a universal solvent, but simply as one tool among many. Not only in health care, when we implement AI to run a conclusion, we need to be very careful about the results.

 

https://blog.pokitdok.com/ai-in-healthcare/index.html

https://www.beckershospitalreview.com/artificial-intelligence/6-ways-ai-is-changing-healthcare.html

Zachary C. Lipton and Jacob Steinhardt, “Troubling Trends in Machine Learning Scholarship,” ArXiv:1807.03341 [Cs, Stat], July 9, 2018.

Gary Marcus, “Deep Learning: A Critical Appraisal,” ArXiv.Org, January 2, 2018.

Using AI to Enhance Education

Can artificial intelligence help to produce real intelligence in the classroom? Can deep-learning algorithms produce deep-learning students? How can these technologies best be implemented and integrated into our education system without putting too much pressure on teachers, students, or the technologies themselves?

These are the types of questions I’m seeking to answer with my upcoming research project.

According to Pew Research data, students in the United States rank near the middle of the pack in math, science, and reading, and are below many other industrialized nations in those categories. According to the 2015 study, among 71 participating countries, the US ranked 38th in math and 24th in science (Desilver, 2017).

In an attempt to help counter this disappointing educational mediocrity, I intend to explore several different aspects of AI and machine learning to discern how these readily available technologies could be utilized effectively in schools.

Based on preliminary research, there are a few elements I plan to emphasize in my writing.

According to Wagner (2018) in a blog on GettingSmart, implementing AI in schools could result in 5 major shifts within schools:

  1. Stand and Deliver Instruction —>  Facilitation and Coaching
  2. Developers of Content —> Developers of Learning Experiences
  3. Siloed Classrooms —> Virtual Social Networks
  4. Textbooks and Set Curriculum —> Blended Courses and Customized Design
  5. Hierarchical Top-Down Network —> Lateral Virtual Global Network

The primary takeaway I gathered from this article was that the role of teaching will need to change alongside these powerful AI technologies. The traditional “sage on a stage” model will give way to a “guide on the side” facilitator role. Classes will become increasingly interdisciplinary and teachers will need to focus on emphasizing media literacy and creating engaging ways of delivering class material across different mediums. Additionally, since AI will be able to automate many basic tasks like grading, tutoring support, and information processing, teachers may need to find a way to supplement some of those automated tasks with personal feedback and support for their students.

The drawback to this notion arises from the Digital Divide, or the older generation of teachers (some still relatively young!) who were not educated– or taught how to teach– using this model or these types of technology. Many teachers in the public school system already feel overwhelmed by the amount of trainings they’re expected to participate in each year, especially trainings that center around technology and teaching style (Stapp, 2018). These are difficult skills to adopt for people who were not direct products of the Digital Age, and especially so when considering the lack of time and other critical factors that play into a teacher’s busy schedule.

Sourced from CDAccess.com

Furthermore, I hope to explore the idea of individualized, adaptive learning programs. With the exception of a few early-learning games that are cartoon-ified and fun (like Schoolhouse Rock, which I loved as a child and still remember playing to this day), many module-based online learning platforms fail to engage students or achieve their desired outcomes because the material is still presented as it might be on a worksheet or set of lecture slides.

Why do we stick to this outdated method of delivery in the classroom, when the games that kids are playing at home (or in their pockets) are infinitely more fun and engaging? For example, the Assassin’s Creed franchise has been around for over a decade, and while the actual objectives can be a bit dark and gory, the entire premise of the game is traveling back to different (historically accurate) time periods and exploring their cities and culture to solve mysteries and track down your targets. Each game covers a different historical era, such as the Ottoman Empire, Ancient Rome, Industrial London, the Italian Renaissance, Revolutionary America, and many more. Simply by playing these games and achieving their objectives, kids can gain a deeper understanding for the culture, architecture, clothes, events, and main characters of these important time periods in our world’s history.

Sourced from Reddit

I have no doubt that similar styles of games could be created for classroom use, that rely on instant gratification, objective-based progression, and even an adaptive CPU that gets harder or easier based on a student’s performance, while still delivering the stunning graphics, dynamic gameplay, and customizable features that kids have come to expect.

The notion of gamified learning is often met with resistance or labeled as “edu-tainment,” but the fact of the matter is that we now live in a society that completely revolves around entertainment. Our phones are always buzzing, social media feeds are always scrolling, TVs are always flashing in the background, headphones are always in, games and events are always being covered, and we recently elected a reality television star as President of the United States. Trying to counter that pervasive entertainment-based lifestyle by teaching children exclusively “the old fashioned way” is doing them a disservice. Of course a balance must be struck between screen-based learning and interpersonal interaction, but at the moment the screen-based learning being implemented is often inefficient and and disengaging for students. If we could responsibly harness the technology that drives the rest of our daily entertainment wants and needs, I think we would see those aforementioned mediocre educational rankings for the United States begin to rise.

 

 

Works Cited

Desilver, D. (2017, February 15). U.S. students’ academic achievement still lags that of their peers in many other countries. Pew Research Center. Retrieved from https://www.pewresearch.org/fact-tank/2017/02/15/u-s-students-internationally-math-science/

Stapp, D. (2018, March). Technology in Secondary Education [Phone].

Wagner, K. (2018, January 15). A blended environment: The future of AI and education. Getting Smart. Retrieved from http://www.gettingsmart.com/2018/01/a-blended-environment-the-future-of-ai-and-education/

Machine Learning: Clarification is Needed

AI is a trending word both in the academic world and in our daily life, but it still remains to be a huge blackbox that people with no science background can barely understand. Among all the AI technologies, machine learning is a method of data analysis that recognizes patterns and automates analytical model building. Machine learning is a great tool that helps researchers to deal with an incredibly large amount of data. However, Lipton and Steinhardt pointed out some trending problems present in machine learning scholarship. The four problems, including the use of mathiness and the misuse of language, that they focus on in their paper can be pertinent explanations for the misunderstanding in this field. In addition, it is noticeable that some of the papers involve a huge amount of computing resources. These researches are difficult to reproduce and verify, which has the potential to bring about the Matthew effect and the monopoly of the academic research. Questions might be asked: Is it necessary to use so many computing resources for machine learning? And how can we get meaningful results from data? Clarification is needed here in terms of the process of machine learning.

At the beginning of this semester, we read articles that give more detailed information about AI/ML from the technical perspective, which helps us go deeper into the applications that we use every day on our phone or websites. But I have to admit that it is hard for me to understand the whole procedures inside the technologies that we talked about in class. For example, I have not figured out convolutional neural networks in facial recognition technology yet. However, I realized that learning the fundamental design principles of technology equips me with the ability to identify the problems in it. For instance, bias facial recognition probably come from biased data. This brings my attention to the stage of data preparation and allows me to think more about what I can do.

All through this semester, we have discussed major ethical and political issues related to artificial intelligence including biased data, privacy, attention manipulation and the loss of human agency. In order to figure out what is going wrong with one specific technology, the first step is to understand the architecture and algorithm of this blackbox. The more clarification there is, the more we can do.

References

 

 

Final Thoughts: Siri and Machine Learning

As the semester begins to close, I am reflecting on some topics that interest me

MACHINE LEARNING AND SIRI

In the Machine Learning piece, Martin and Jurafsky, helped me gather some fundamentals on my inquiries. When we de-black box this, we can see there is no perfect way to translate something, especially how I mentioned before that the “perfect” translation doesn’t exist in all of the same locations, since not all language is “universal”. They say, “Technical texts are not translated in the same way as literary texts. A specific text concerns a world that is remote from the world of the reader in the target language. The translator has to choose between saying close to the original text or making use of paraphrasing to ensure comprehension. The tone and style of the text are highly subjective” (Machine Learning, pg. 19). This got me thinking, How, can we trust machine translation or google translate so much if it is impossible to gain 100% accuracy? Where does this trust reside?

( week 7 )

From the Wikipedia article this week, Siri was defined as, “the virtual assistant for Apple systems that uses voice queries and natural language user interface to answer questions, make recommendations, and perform actions through requests of the internet. The software adapts to the user’s individual language usages, searches, preferences, etc. The results are personalized” (Wikipedia, Siri, pg. 1). This helped me pieces together what it really is doing and how it’s a personal experience and technology. This made me raise the question of now that I know this and some basic layers to how the voice recognition works, what else can we do with siri? 

I wanted to look for other outside material to help answer my question, since its pretty standard that siri is just basically a voice over platform that we use to speak into in order to receive some type of mediated response or aid to an inquiry. According to writer Todd Haselton, for CNBC, with the competing rise to Google Assistant, we can now talk to Google Assistant, THROUGH your Siri. This new technology is basically Google’s own version of Siri, which is somewhat funny that you can use the competition to open up the other app. When you do this, you must APPROVE the ability to pair Siri with your downloaded google assistant, assuring the same voice over controls. When you’re ready, you say, “Hey Siri, OK GOOGLE” (Haselton, pg.1). This allows the Siri to approve your voice and open up the Google assistant instead.

( week 8 )

________________________

For this week, I wanted to add more readings. One article that stood out in particular was The Verge piece that discussed the ethical side to AI a bit further. They discuss a more European way to enhance the ethical and privacy issues that they face with AI and made a list of 7 things that they feel is the most important and needs work. In their list they say, “human agency and oversight, more safety, securing privacy and data, better transparency for humans to explain the algorithms chosen and implemented, more diversity and fairness, sustainable technology that will promote well-being rather than changing it, and more accountable on restoring privacy and further updates to the technology. (The Verge, pg.1).

For me I found this article really important in relation to my class connections that I have made over the semester. I think the most important role in AI that I have learned in the privacy, and code of ethics that must be implemented in these technologies that have not only given rise to their advancements but their disadvantages too, which of course can be a huge issue.

All of this is helpful in guiding me towards preparing for my final project. I met with Dr. Irvine and we helped narrow down my interests in this issue and how it plays a role in Apple products, like Siri. I took the most interest in our virtual assistant and privacy week work. I think this would be a great topic and issue to look into and gather a socio-technical-historical approach to the technology and incorporate the development, ownership, production, and implementation of Siri, and the future of where it’s going. I appreciate Siri and its ability to aid so much in our every-day lives. In connection to this reading, this would be a great example and issue to pull in my literature review and see what other pieces of literature I can find on this issue and other key features to Siri, from then until now.

A question I have developed is with recent advancements, research, and reports on AI ethics and privacy, where do the boundaries get drawn when we look ahead towards the future? Where do the limitations start and end? What will the future of AI look like with these advancements in place? I look forward to unpacking more answers and insight to these questions in the remainder of the course and final paper.

 

https://www.theverge.com/2019/4/8/18300149/eu-artificial-intelligence-ai-ethical-guidelines-recommendations

 

Haselton, T. (2018, November 21). You can now talk to Google Assistant through Siri on your iPhone here’s how. Retrieved March 12, 2019.

 

Daniel Jurafsky and James H. Martin, Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition, 2nd ed. (Upper Saddle River, N.J: Prentice Hall, 2008).

Deep Learning’s Pros and Cons

Deep learning is essentially a statistical technique for classifying patterns, based on sample data, using neural networks with multiple layers. Deep learning is a subset of machine learning. Deep artificial neural networks are a set of algorithms that have set new records in accuracy for many important problems, such as facial recognition, sound recognition, recommendation systems, natural language processing, etc. Although it has brought much convenience for human beings, people still have a critical appraisal against it. I want to take the facial recognition system as an example.

The facial recognition system is a technology capable of identifying person from a digital image or video. It works by comparing selected facial features from given image with faces within a database. Consequently, in order to make sure that the outcome is accurate, the deep learning works best when thousands, millions, or even billions of examples or data have been collected in the database. However, if the database is not big enough, the validity of the outcome will no doubt be questioned.

In addition, the safety of information can also be doubted in the facial recognition system. If your face can directly lock your phone, your other personal information will become visible to others as well. And others might also use your face to log in your bank account and steal your money. The source of the facial information can easily retrieve from the social media, from any selfies you posted on the internet. You don’t hope that your facial password is always exposed to anyone else, right?

Also, the feasibility of the facial recognition system can be doubted. Deep learning presumes a large stable world, while in real world, it can be constantly changing. If the system makes some mistake, while people trust this system, the outcome might be serious. The source of the database for the facial recognition system is difficult as well. The facial information is hard to be captured online. If the facial information is captured offline by police and used online, when the database was hacked, or destroyed, the outcome can be so dangerous and horrible.

In the end, this technology is also related to moral issues, and that is why mosaic was used in the news and other videos. Face, as one of the people’s personal information, has the right of privacy like others. The facial recognition system no doubt violates this right and will raise moral debate among people, although it is quite interesting and convenient.

 

Questions/Topics:

I am quite interested in the information safety of the cloud system, this is more significant in the public cloud and community cloud. Since the information and resource can be shared in cloud through the internet, does it mean that information will not be private? What if a company wants to enjoy the benefits of the cloud system as well as keep the information secret to a certain group of people without being violated by others?

 

Reference:

  • Agre, P. (1997). Computation and human experience. Cambridge; Cambridge University Press.
  • Thorn, P. (2015). Nick Bostrom: Superintelligence: Paths, Dangers, Strategies. Minds and Machines25(3), 285–289. https://doi.org/10.1007/s11023-015-9377-7

Challenges of (not)Fitting into the Deep Learning Model

Gary Marcus offers a very detailed critique of the field of AI in “Deep Learning: A Critical Appraisal”. What caught my attention regarding this article was the fact that it was presented as an intentional self-introspective snapshot of the current state of deep learning. It looks not only at how much as been accomplished but also how it has failed and what that presents as different approaches to deep learning in the future.

He says, “deep learning currently lacks a mechanism for learning abstractions through explicit, verbal definition, and works best when there are thousands, millions or even billions of training examples, as in DeepMind’s work on board games and Atari. As Brenden Lake and his colleagues have recently emphasized in a series of papers, humans are far more efficient in learning complex rules than deep learning systems are (Lake, Salakhutdinov, & Tenenbaum, 2015; Lake, Ullman, Tenenbaum, & Gershman, 2016).” (p. 7)

As we have mentioned many times before, deep learning struggles to offer outputs that accurately reflect complex human concepts that are difficult to represent as computational, as a set of yes and no answers. It can go from translation of languages (I never get tired of using this very convenient example) to more abstract concepts such as justice.

Referring to my personal favorite example of open-ended natural language, Marcus says, “In a problem like that, deep learning becomes a square peg slammed into a round hole, a crude approximation when there must be a solution elsewhere.” (p. 15)

Another observation present in Marcus’ analysis refers to the approach of a real world taken as a ‘set in stone’ reality: “deep learning presumes a largely stable world, in ways that may be problematic: The logic of deep learning is such that it is likely to work best in highly stable worlds, like the board game Go, which has unvarying rules, and less well in systems such as politics and economics that are constantly changing.” (p. 13)

Not only the world and our knowledge of it is constantly changing, but our representation of that reality through data is most of the times inaccurate at best, skewed at worst. To what extent and what are the different ways in which we can see the impact of such flawed outputs? Sternberg presents two aspects to take into consideration:

  • What exists in the data might be a partial representation of reality:

Even that partial representation might not be entirely accurate. For example, the famous case of Kodak’s film being unable to efficiently capture non-white tones of skin is also present in facial recognition systems. Other controversial cases were those that mistook pictures of Asians as ‘blinking’ and identified black people as gorillas. The social cost of a mistake in any AI system being used by the police for decision-making is high and more likely to present results that are less accurate with minorities since they were underrepresented and misrepresented in the data-set: “This also calls for transparency regarding representation within the data-set, especially when it is human data, and for the development of tests for accuracy across groups” (Sternberg 2018, October 8).

  • Even if the data does represent reality quite truthfully, our social reality is not a perfectly-balanced and desired state that calls for perpetuation: 

As an example, gender and racial biases present in binary terminology is, after all, based on statistics present in the off-line world, well-documented in history. However, here Sternberg offers an optimistic idea, “our social reality is not a perfectly-balanced and desired state that calls for perpetuation”. This meaning that we are putting so much value on the data and deep learning in this process, giving it a deterministic characteristic, when in reality, these are not the ideas, concepts and human values we should be preserving or basing our technology on.

In regards to that, Sternberg criticizes the absolute faith in the outcome of these systems regarding them as more objective than humans: “Sexism, gender inequality and lack of fairness arise from the implementation of such biases in automation tools that will replicate them as if they were laws of nature, thus preserving unequal gender-relations, and limiting one’s capability of stepping outside their pre-defined social limits” (Sternberg 2018, October 8).

Marcus’ premise agrees with Sternberg, focusing more on the problem of thinking about deep learning as the only tool available to understanding and digitalizing the world when, in reality, this tool might not fit every problem we want to fix: “the real problem lies in misunderstanding what deep learning is, and is not, good for. The technique excels at solving closed-end classification problems… And some problems cannot, given real world limitations, be thought of as classification problems at all.” (p. 15)

These ideas are not entirely new to me. However, what I found the most thought-provoking about Marcus’ article is the proposal to see deep learning as this set box in which every human problem/though must be filtered through, but to understand that we have to develop other hybrid ways in which we can analyze these problems beyond classifications instead of trying to make the square leg fit into the round hole.

 

References: