12/12/2017 (Finals Week) – Student Presentations II

In our last session (Finals Week) we will do part II of student presentations. Blog comments may address one or more of the excerpts assigned if you are not already assigned to a specific book for commenting in this session. The schedule of presented readings is as follows:


7:15 Annie Jacobsen, The Pentagon’s Brain: An Uncensored History of DARPA, America’s Top-Secret Military Research Agency (2016)

7:30 Robert Kyncl, Streampunks: Youtube and the Rebels Remaking Media (2017)

7:45 Brad Stone, The Upstarts: How Uber, Airbnb, and the Killer Companies of the New Silicon Valley Are Changing the World (2017)

8:00 Robert Cialdini, Pre-Suasion: A Revolutionary New Way to Influence and Persuade (2016)

8:15 Andrew W. Lo, Adaptive Markets: Financial Evolution at the Speed of Thought (2017)

8:30 BREAK

8:35 Pedro Domingos, The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World (2015) (The page groupings are as follows — some page numbers are cut off: Pages 24 – 26, Pages 52 – 53, Pages 141 – 142, Pages 174 – 175, Pages 267 – 270, & Pages 288 – 289)

8:50 Hector J. Levesque, Common Sense, the Turing Test, and the Quest for Real AI: Reflections on Natural and Artificial Intelligence (2017)

9:05 Marina J. Lostetter, Noumenon (2017) [fiction] (Focus simply on the parts of these passages discussing I.C.C., an apparently conscious computer who leads an envoy of interstellar generation ships from Earth to a distant star named LQ Pyxidis.)

9:20 Yuval Noah Harari, Homo Deus: A Brief History of Tomorrow (2017)




Filed under Class sessions

15 responses to “12/12/2017 (Finals Week) – Student Presentations II

  1. tamara2205

    Reading the excerpt from “Upstarts” on Uber’s early battles for market dominance, I was struck by the machinery of CEO’s, lawyers and lobbyists that hindered or helped the startup rise over regulation controversies. One thing that particularly caught my attention was “Travis’s Law,” implicitly passed by Travis Kalanick. Catchy and irreverent as any part of a good startup fable, Travis’s Law dictates that a government responsive to its people could be pressured by them to accommodate any service that is markedly better than the status quo. In the context of Uber, this meant “mobilizing their user base and directing their passion toward elected officials” (194) in order to circumvent the carpooling regulations that made their business illegal.

    While the implications sound quite democratic and user-friendly, I am not sure that enlisting customer help should have that much weight in deciding the safety and legality of a new technology — or at least not without careful scrutiny of the service itself by ethics professionals. Unlike legal committees, whose job is to deliberately examine a technology’s potential benefits and safety concerns, users are more likely to be blinded by the benefits only and thus request legal change. A mother should not let her child eat 10 boxes of Cheerios just because he thinks it would make his life more fun, as children often lack the necessary framework to reason about their safety. After all, Stone explains that “ridesharing” was a term used to eliminate concerns about getting into strangers’ cars in order to “fit the idea within legal protections” (196). It was thus branding and marketing that got users to embrace the service and push for its legality, not their dispassionate judgment. And while it worked out well for Uber, generalizing this principle might not always have such bright legal and safety consequences.

  2. RE: “Presuasion” by Robert Cialdini

    The leading case that the author makes throughout the passage is that the moments before making an argument or the actions before you ask a favor are vital; they can affect how influential your message is or how responsive someone is to performing a favor for you.

    However, to me, Cialdini’s “Presuasion” passage, was less about persuasion specifically and more about tactics to get people to like and trust you. He even names “liking” as one of his key pillars of influence and the idea of “liking” and trust seeps into all other aspects that he lists: reciprocation, social proof, authority, scarcity, and consistency (pg. 1). I see how building camaraderie follows naturally into persuasion – I am much more likely to do a favor for someone I feel an affinity to.

    The work reminded me of Dale Carnegie’s “How to Win Friends and Influence People” (1937), which also discusses subtle psychological tactics one can take to increase the goodwill others have towards them. Both Cialdini and Carnegie are in the business of teaching their pupils how to convince someone of something most efficiently, be it through flattery or reciprocity. Both can lay out in digestible words what makes some people more successful in influencing others and some people less effective.

    Perhaps later on in “Presuasion,” the author will discuss how these how his argument is unique from that of Carnegie’s, which he published 80 years ago. But now, through this given passage, I’m struggling to see how his advice and tactics are any different from the ones that Carnegie laid out, other than just that Cialdini’s is an update with modern examples of Carnegie’s seminal work.

  3. Reading “The Upstarts”, I am struck most of all by the ability of consumers to break down entrenched economic and political monopolies. p.191 details taxicab regulations that do little more than harm Uber, which is of little surprise. Taxicab companies are granted state-blessed monopolies to engage in rent-seeking behaviour, thanks to the medallion system for hailable cabs and tight regulation for bookable taxis. Those monopolies then lobby for their continued existence, via pandering over safety and regulation (for example, London’s decision to shutter Uber on a technicality over criminal checks: https://goo.gl/s2LYyV) and appeals to the sunk-cost fallacy (medallions often cost hundreds of thousands of dollars).

    The historic model for trying to break those monopolies is to lobby Congress, as Kalanick tries (and fails spectacularly) to do in DC. But such a model is flawed: not all CEOs are especially good at such appearances (see Kalanick); political lobbying is expensive and uncertain; it likely enhances the total level of corruption in Congress; and, critically, startups that thrive in free marketplaces will never defeat taxicabs whose very existence is dependent on legislators.

    “Travis’ law” as described on p.195 does away with this lobbying, replacing it with a simpler one: if a product is better, people will demand it be permitted. But this misses the critical reasons Uber broke the taxicab monopoly: first, its accessibility via smartphone made it a viral product; and second, Uber actively encouraged one-click emails to Congressmen before critical votes.

    Uber, as demonstrated by its enormous popularity and success, offers customers a better product at a lower price than its competition. Kalanick deserves credit for pioneering a growth model that leveraged the power of individual consumers as political change-makers, rather than trying to lobby as a company, enhancing corruption, and inevitably losing to inefficient and politically connected adversaries.

  4. Chapter four of Annie Jacobsen’s /The Pentagon’s Brain: An Uncensored History of DARPA, America’s Top-Secret Military Research Agency/ details DARPA’s (Defense Advanced Research Projects Agency) approach to autonomous devices, vehicles, and weapons. The bulk of this chapter has to do with DARPA’s move towards “biosystems, biomimetics, and biohybrids” – “machines that work like real animals “scientists build mechanical systems to imitate creatures from the natural world” (410). There have some incredibly cool projects, for example at Cal: “Scientists implanted electrodes into the brain and wings of a 2-centimenter-long beetle and sewed a radio receiver onto its back. By remotely delivering electrical pulses to the beetle’s brain, they were able to start and stop the beating of the beetle’s wings, thereby steering and controlling the insect in flight” (412). This kind of cyborg insect may sound sci-fi and futuristic, however, on principle, robbing animals of autonomy to aid in human endeavors is an age-old idea. Jacobsen, I think correctly, compares the situation to training German shepards to track Vietcong soldiers during the Vietnam War (410).

    Her greatest insight is that there is a blurry line between devices, vehicles, and weapons – what starts as an unmanned aerial listening (an MAV – micro-air vehicle) device can quickly be outfitted to bear weight/transport goods, and then be used to execute strikes. It would have been helpful here to use some historical examples of this phenomenon – while the slippery slope notion seems reasonable in theory, we also know of many of examples of DoD-funded projects that become don’t necessarily ratchet-up right away – they instead diffuse. See for instance, ARPANET, which began as a DARPA project to share Cold War-related research, and then became the invaluable civilian commodity known as the Internet. Perhaps in practice DARPA projects diffuse and intensify simultaneously.

  5. rhwang2

    In this excerpt of Adaptive Markets, the author presents the Adaptive Market Hypothesis, which essentially claims that our features and behaviors “are shaped by the forces of evolution” and in particular our environments. To illustrate his hypothesis, he presents the example of the tribble, an imaginary animal which must decide whether or not to mate reproduce in the valley or plateau, only one of which will be too inhospitable to reproduce. In this scenario, which environment is inhospitable is decided on a set random probability, and so the types of tribbles that eventually dominate the population are those that match the random probability in deciding where to reproduce. If the valley is the optimal choice 70% of the time (implying that the plateau is 30% the time) then the tribbles that chooses the valley 70% of the time will dominate the population.
    Lo explains that many of humankind’s irrationalities are not really irrationalities but more like optimizations for an outdated environment. For better and for worse, evolution has largely halted, and unfortunately our current environment continues to develop further and further away from the environment we are actually optimized for. The likely result is then more irrational behavior then before. I wonder then if we will eventually reach a point when the incompatibility with our modern environment becomes too significant to ignore. Although internet addiction is seemingly innocuous, consider the enormous amount of wasted productivity produced by the exploitation of human biases. What if, for example, our psychological tendencies for cooperation vs. competition are too outdated to prevent mutually assured destruction or severe irreversible environmental degradation?

  6. Caroline Cin-kay Ho

    In Noumenon, the apparently conscious friendly AI ICC composes a poem “without prompting,” causing crew member Vega Hansen to exclaim that it has “exercised pure free will” and done “something beyond [its] programming” (213). This action initially seems to be fundamentally different from ICC’s choice to short out its system (analogous to committing suicide); while the latter was done to achieve its “one goal in life: see to the mission’s success… keeping everyone alive and the society stable” (159), the former served as “an outlet” (214) and (according to ICC) did nothing to increase its effectiveness as a crew member (214). However (putting aside the question of free will for now), I would argue that poem-writing might not be necessarily beyond ICC’s programming. As Tegmark notes in Life 3.0, AI will necessarily develop certain subgoals based on its ultimate goal (263), such as “self-preservation and resource acquisition” (265). We do see some evidence of ICC having the subgoal of self-preservation, as it reports feeling “sick,” with many “bells and whistles and warning lights [springing] to life” (159) as it contemplates shorting its system as a last resort. Likewise, because ICC’s ultimate goal has to do with keeping society stable, it might have developed the sub-goal of understanding humans and their subjective experiences to the fullest – and in the process developed human-like patterns of thought. Just as the human “brain can rebel against [its] genes by choosing to use contraceptives” (Tegmark 256), choosing to satisfy its feeling of lust rather than work toward its goal of replication, ICC may have done the apparently pointless action of composing a poem in contravention of its ultimate goal (hence leading to its confusion over its action) but in accordance with its sub-goal.

  7. In the excerpt from “Pre-Suasion”, Robert Cialdini explains the “psychological force” reciprocation holds to “influence recipients towards assent.” (2) The discussed rule for reciprocation says “that those who have given benefits to us are entitled to benefits from us in return.” (4)

    The psychological influence of reciprocity is hardly surprising – a similar rule is present in many faiths through the Golden Rule: “do to others what you want them to do to you.” The major difference in paradigms of reciprocity is the concept of “entitlement”, since the Golden Rule suggests that others do not owe anything to those who treat them well – that peoples’ obligation to treat others the way they would like to be treated remains independent from the way they may actually be treated. By the Golden Rule, for example, one offers someone a gift because they, too, would appreciate a gift. By Cialdini’s rule, one offers a gift and holds the entitlement of receiving a gift in return.

    Cialdini’s reciprocity is thus transactional – giving someone something and taking advantage of conditioning to the Golden Rule to receive a favorable outcome. However, this rule becomes ineffective if the obligation created is unnatural. Take, for example, a strategy from the “Krishna Society, where a robed Krishna would walk up to a person and give them a gift…[which] when people attempted return, [was] refused,” (Cialdini) requesting a donation instead. This was ineffective as reciprocity was used “to create obligations that didn’t exist naturally – exploiting it so only they benefited,” from the rule (Cialdini). Hence, a challenge in using reciprocity to influence is understanding what constitutes a “benefit” to others. Cialdini’s response to this problem is suggesting a reciprocity-optimizing triumvirate including “customization” – “when a first favor is customized to the needs, preferences, or current circumstances of the recipient.” (11)

    Cialdini, Robert B. “The Power of Persuasion (SSIR).” Stanford Social Innovation Review, Stanford University , 2003, ssir.org/articles/entry/the_power_of_persuasion.

  8. Divya Siddarth

    Despite Pedro Domingos’ evangelizing tone, the Master Algorithm, sets out, across these excerpts, a good introduction to technical and non-technical aspects of machine learning. I found the separation into groups of thought – based on my understanding: symbologists (logic / deduction), connectionists (backpropagation), evolutionaries (natural selection algorithms), Bayesians (probabilistic inference), and analogizers (similarity-based learning) (52-53, 141-142) – somewhat reductionist, but appreciated the look at how different schools of thought deal with fundamental issues in creating a generalized algorithm. However, even through the quick overview presented in the excerpts, one can certainly find evidence that Domingos falls heavily on the side of not only expecting a world in which a ‘Master Algorithm’ will someday be created, but explicitly welcoming such a world. Coming from our readings this quarter, particularly Weapons of Math Destruction and Life 3.0, I would definitely caution Domingos into examining different possibilities than the one he has laid out (in which the ‘Master Algorithm’ is entirely controlled by humans, and exists to perfectly make our collective lies better, safer, and more fulfilling (267-270)). He does not discuss AI risk, and seems convinced that this ‘Master Algorithm’ will lead to only progress, perhaps ‘the most amazing years in the life of planet Earth’. Further, when confronted with questions of whether such an algorithm will lead to a stratification within human society and a deepening of inequality, he seems to dismiss those who raise these objections as subject to a ‘failure of imagination’ (289). I certainly acknowledge that easy dismissal may be fleshed out in other parts of the book I did not have a chance to read. However, I would say if this not the case, perhaps the failure of imagination is on the side of Domingos’ Master Algorithmic utopia, rather than on the side of the dissenters.

  9. Bradley Richard Knox

    I found this excerpt of Homo Deus incredibly interesting, and in some ways very realistic in the not-so-distant future. Much of the section on a machine or other intelligence system knowing your medical history and essentially making your decisions for you reminded me of Santosh’s presentation on the future of medicine, and how it is becoming increasingly individualized and in control of the user. This excerpt takes that idea to the next level, where not only is your medical journey increasingly individualized, but every decision that is being made about it is done so using a highly intelligent system.

    However, I do think that there is danger in handing over all of our decision making to machines that seem to always know what is best for us. I believe that there is a distinction between an intelligence system knowing us better than we do, and knowing what is best for us better than we do, and I feel as though this author is blurring that line. I don’t doubt that something will eventually be able to tell me that some decision that I will make is better than another, but that may take away much of what it actually means to be human, and I think that people will resist this possible loss of free will to make their own mistakes and learn from them. I don’t know what the author’s main thesis is for the book, but I just wanted to push back a little bit in saying that there are definitely some decisions that I wish that I didn’t have to make on my own, but also that it seems like a dangerous road to travel down if you believe as I do that making mistakes and learning from them are crucial to the human experience.

  10. SM201

    In this excerpt from The Upstarts, Uber is portrayed as a seemingly immovable force. Despite Kalanick’s disastrous testimony against government regulation, the company was still able to leverage the voice of its user base to influence governmental policy, clearing the way to disrupt decades- or centuries-old industries; sometimes even against the wishes of elected officials. (194)

    Specifically, councilman Jim Graham states “I don’t want this city to be [all] Uber. I really don’t. Because there’s too much of a history of our taxicab industry.” (193) Many may characterize Graham as a Luddite; certainly, culture is bound to change over time. Yet, we do have protections for concepts of cultural significance – for example, UNESCO World Heritage Sites. Some may even argue that the sight of yellow taxicabs in New York or D.C. is just as worthy of protection. This points to a larger issue (one we saw in Life 3.0 as well): How does one decide what “historical” things should be kept? Moreover, can you regulate the rise of new companies in a way that promotes better products, without damaging these vestiges?

    Given that this excerpt is historical narrative, it’s challenging to identify Stone’s perspective on this topic. From the perspective of Kalanick or other tech giants, the preferred approach is to give the general public full control over determining which product to use, even if it means the upheaval of old industries.

    Perhaps Kalanick’s approach could be improved by borrowing regulatory tactics from other industries. Take healthcare. When a new drug is developed, even if the population is clamoring for it, there is still a set of regulations that Big Pharma must abide by (even under FDA Fast-Track status). This could be a first step towards striking a compromise between Graham’s regulation and Kalanick’s desire for revolution.

  11. Julia Thompson

    In this excerpt from Common Sense, the Turing Test, and the Quest for Real AI, Levesque discusses the possibility of knowledge-based computer systems. He concludes through using the ideas of Gottfried Leibniz that symbol processing abilities are necessary for any knowledge based system. Math is done through symbols, as numbers are abstract concepts. Therefore, the representation of those numbers can change. Most ideas, in fact, are similarly abstract, so we use words, images, and a variety of other symbols to convey ideas. A knowledge-based system would have expressions stored in its memory as its knowledge.
    However, storage in memory does not seem adequate for something to be described as knowledge. A common definition of knowledge is justified true belief. While that definition has flaws, it seems applicable here. In this case, the justification would likely be because a human had given the system the information. Unless the system is capable of critical thinking and evaluating human sources, the system would not necessarily be able to distinguish a human who should be trusted as a justification and one who should not. This leads to problems with the truth of the information that the system has. Even if its symbol processing is adequate and can lead to valid conclusions, these conclusions could also be false if the information they are drawn from is. If the input information is either false or misleading, this system would still treat it as knowledge, which could cause a variety of problems. Further complicating this is the idea of belief, which implies conscious experience. This would make such a system significantly different than just a symbol processor, similar to a calculator. Storage in memory should not be sufficient for information to be in a knowledge base.

  12. Two parts of the Streampunks excerpt stood out to me
    – That competition in media has shifted from shelf space to time. Thus, “attention is the currency of our age” (23).
    – That we spend so much time with video, hidden in the corners of our day
    Both observations echo Wu’s in Attention Merchants. Wu spends several chapters describing how mass media, which peaked in the 50s, has decentralized into cable networks, websites, and now personally-tailored news feeds. He also discusses how attention merchants take advantage of our need to always pay attention to something, taking advantage of every moment we are bored, in between tasks, or feeling restless.
    Where Wu focuses on all kinds of media, Kyncl singles out video, and specifically video streaming, as the most import form of media. It’s hard to deny his claim when we look at the world around us. Facebook, for instance, now prioritizes video and the “share” feature in feeds, rather than its traditional pattern of user-published posts. Not to mention the class of apps which center entirely around sharing video – Snapchat, and formerly Vine as chief examples.
    Wu decries this environment as a place where we, as media consumers, are overwhelmed and unable to devote our attention thoughtfully. He worries that the fragmentation of media makes it increasingly hard to sustain a democracy, where citizens need to discuss and make decisions together. Kyncl seems far more optimistic. He looks at this freedom granted to consumers (which Wu acknowledges as well when discussing Netflix) and also applies it to content creators. People are able to express their ideas and share their interests through a powerful medium once guarded by powerful groups – now, anyone can find a community centered around videos they are interested in, whether that’s news, physics videos, or skits.

  13. Alex Gill

    In this excerpt from “Streampunks” the author notes that YouTube, which pays creators shares of advertising revenue, has “helped bolster the bottom lines of traditional media companies and music videos,” citing the fact that they’ve paid out $3 billion to the music industry as proof (23). While I certainly support YouTube’s commitment to paying content creators – an unfortunately rare business model – I question how much YouTube has “bolstered” these media companies considering the amount of piracy that occurs through their platform.
    Stream-ripping is an incredibly popular way for people on the internet to get free downloaded copies of audio from nearly any video on YouTube. In fact, the Financial Times reported last year that nearly half of 16 to 24-year-olds use stream-ripping software to illegally copy content. The popularity of this form of piracy makes me question if the $3 billion that YouTube pays to the music industry is actually worth the amount of money they inevitably lose because of stream-ripping.
    The music industry has seen a major shift in the past 5 years as more and more people have begun to adopt streaming platforms such as Spotify to listen to music. Although Spotify is certainly controversial in terms of the amount of money it pays out to artists per stream, it does provide a somewhat-cheap alternative to piracy while hosting a variety of independent and small-scale artists. Therefore, it seems to me that Spotify can act as a sort of “best of both worlds” platform that offers the unlimited music that customers want without such a high risk of piracy for artists.

    FT Article: https://www.ft.com/content/d31ff954-793a-11e6-a0c6-39e2633162d5

  14. Meg Verity Elli Saunders

    Cialdini’s examples demonstrating the power of reciprocity are remarkable, particularly in the instance of Abu Jandal. In this example, Osama Bin Laden’s former bodyguard, after repeatedly providing no responses to his interrogators, on receiving sugar free cookies with his tea, obligingly “provided extensive data on Al Qaeda operations, as well as the names of seven of the 9/11 hijackers.”(14 of excerpt) Cialdini attributes this effect to the fact that the act of bringing the sugar-free cookies to the diabetic prisoner is characterised by the three conditions that he defines as being optimal for inducing reciprocity. These conditions are that the initial gift must be “experienced as meaningful, unexpected, and customized.”(8 of excerpt) What stands out to me in this quote is Cialdini’s subtle distinction that the gift is experienced this way. The commonality among these features is that they all create the impression that the giver has gone to great lengths to provide the perfect gift for a person, making that person feel morally obliged to return the favour.

    Perhaps the reason why this technique of persuasion was so effective in the above example is that to Abu Jandal in particular, this act of kindness was so unexpected. In reality, the cost and effort of obtaining sugar free cookies undoubtably would have been negligible to Jandal’s interrogators. However, Cialdini describes how Jandal’s previous responses to questioning had been merely “screeds against the ways of the West.”(13 of excerpt) What this shows is that Jandal’s opinion of Western culture was decidedly low. Therefore, when they did this tiny thing for him, it would have been magnified in his perception because it contradicted, albeit on a very small scale, his preconceptions of how Westerners are, making this gift seem more meaningful to him.

  15. In this passage from Pre-Suasion, the author discusses six concepts which empower the principles of human social influence: reciprocation, liking, social proof, authority, scarcity, and consistency. She argues that these are the general principles leading to acceptance of a view or belief.

    However, the idea of acceptance seems convoluted in light of foreign influence on the presidential election and social justice advocates in general. Rather than pushing for acceptance of their own view, foreign advertisements for presidential candidates facilitate the perspective of another person. While Trump could change his position on certain topics, these external parties are left to leverage here principles of human social influence to sway their audience. Thus, the principles that she discusses can be transferred to those advocating for others. As there were many more advertisements from random sources than assertions by Trump, the ability to identify and evaluate such principles becomes even more important for the public.

    Out of the six concepts the author discusses, she specifically goes into depth about reciprocation – the idea of owing another something when a good deed is done. While advertisements representing services or deeds during the election are rarely particularly meaningful and unexpected, she emphasizes the importance of customization. This is particularly relevant for foreign campaign advertisements, given the ease of targeting custom audiences on platforms such as Facebook. By determining what positions a user cares most about provided their online data, advertisers can warp and lie about a candidates position on the topic, skewing it towards the user’s preference. If someone is told that Trump is addressing all the problems they care about in favorable ways, then they are probably more likely to vote for him. This calls for a greater self-awareness in addition to awareness of these tactics to properly evaluate the media.

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s