Interesting Stuff from the Internet
This page contains content for the Artificial Life course (2023). Although it is technically a PhD course in the Math department, there are no hard requisites, and it is suitable and even intended for students in Biology, Physics, and Computer Science. The course aims to give an overview of the field of Artificial Life, revisiting some of the classic theoretical results and models. In the latter half of the course, emphasis will be put on the description and implementations of more recent models in code. On this page, you will find recordings and slides of the lectures and any other material used in class.
Occam's Razor (Law of Parsimony):
Principle: Among competing hypotheses, the one with the fewest assumptions should be selected.
Explanation: Simpler explanations are generally better than complex ones because they are easier to test and falsify. This does not mean the simplest explanation is always correct, but it is a useful heuristic to prioritize simplicity.
Hitchens' Razor:
Principle: What can be asserted without evidence can also be dismissed without evidence.
Explanation: This razor emphasizes the importance of evidence in making claims. If someone makes a claim without presenting evidence, there is no obligation to take it seriously or provide a counter-argument.
Hanlon's Razor:
Principle: Never attribute to malice that which is adequately explained by stupidity (or incompetence).
Explanation: People often jump to conclusions about others' malicious intent when simple incompetence or ignorance could be the cause. This razor encourages giving others the benefit of the doubt and considering simpler, non-malicious explanations first.
Alder's Razor (Newton's Flaming Laser Sword):
Principle: If something cannot be settled by experiment or observation, then it is not worth debating.
Explanation: This principle, attributed to physicist Mike Alder, suggests that only empirical evidence-based arguments are meaningful in scientific and practical discussions. Abstract debates without empirical grounding are less useful.
Sagan Standard (Extraordinary claims require extraordinary evidence):
Principle: The more extraordinary the claim, the stronger the evidence needed to support it.
Explanation: Popularized by Carl Sagan, this principle asserts that if someone makes a highly unusual or surprising claim, they should provide equally compelling evidence to substantiate it.
Hume's Guillotine (Is-Ought Problem):
Principle: One cannot derive an "ought" from an "is."
Explanation: This principle, derived from David Hume's work, highlights the difficulty of deriving prescriptive statements (what should be) directly from descriptive statements (what is). Moral and ethical prescriptions require more than just factual descriptions.
Grice's Razor:
Principle: Do not assume more meaning than necessary in what people say.
Explanation: Named after philosopher H.P. Grice, this principle suggests interpreting others' statements in the simplest and most straightforward way possible, rather than reading into them unnecessary complexity or hidden meanings.
There are four main areas to consider when writing an effective prompt. You don’t need to use all four, but using a few will help!
Persona
Task
Context
Format
Here is an example of a prompt using all four areas that could work well in Gmail and Google Docs:
You are a Google Cloud program manager. Draft an executive summary email to [persona] based on [details about relevant program docs]. Limit to bullet points.
Here are quick tips to get you started with Gemini for Workspace:
Use natural language. Write as if you’re speaking to another person. Express complete thoughts in full sentences.
Be specific and iterate. Tell Gemini for Workspace what you need it to do (summarize, write, change the tone, create). Provide as much context as possible.
Be concise and avoid complexity. State your request in brief — but specific — language. Avoid jargon.
Make it a conversation. Fine-tune your prompts if the results don’t meet your expectations or if you believe there’s room for improvement. Use follow-up prompts and an iterative process of review and refinement to yield better results.
Snippets from the Newsletters/ Newspapers/ Books
“Because the history of evolution is that life escapes all barriers. Life breaks free. Life expands to new territories. Painfully, perhaps even dangerously. But life finds a way.” ― Michael Crichton, Jurassic Park
The Kuleshov Effect is a film editing (montage) phenomenon named after Soviet filmmaker Lev Kuleshov, who demonstrated that viewers derive more meaning from the interaction of two sequential shots than from a single shot in isolation. The effect highlights how the juxtaposition of images can influence the audience's perception and emotional response, suggesting that meaning in film is constructed through editing.
Key Points
Experiment:
Lev Kuleshov conducted an experiment in the 1910s and 1920s where he used the same shot of actor Ivan Mosjoukine's neutral face and intercut it with various other images (a bowl of soup, a girl in a coffin, a woman on a divan).
Despite the actor's expression remaining the same, audiences interpreted his emotion differently based on the context provided by the adjacent shot (hunger, sorrow, desire).
Significance in Filmmaking:
The Kuleshov Effect underscores the power of editing in film. It demonstrates that viewers' interpretations of a scene are shaped significantly by the sequence of shots.
It suggests that the meaning of a scene is not inherent in individual shots but is created by their combination.
Impact on Film Theory:
The Kuleshov Effect has had a profound influence on film theory, particularly in the field of montage theory, which emphasizes the importance of editing in generating meaning and emotion in cinema.
This concept is foundational to the work of Soviet montage filmmakers like Sergei Eisenstein, who expanded on Kuleshov’s ideas.
Applications:
Filmmakers use the Kuleshov Effect to guide audiences' emotions and thoughts by carefully arranging sequences of images.
It is a tool for storytelling, allowing filmmakers to convey complex ideas and emotions without explicit dialogue or narrative exposition.
Example
Imagine a sequence of three shots:
Shot A: A neutral expression on a man's face.
Shot B: A steaming bowl of soup.
Shot A: The same neutral expression on the man's face.
Audiences might interpret the man as feeling hungry. If Shot B is replaced with an image of a child playing, the audience might perceive the man as feeling affectionate or happy. This change in interpretation, despite the man's expression staying the same, illustrates the Kuleshov Effect.
The Halting Problem is a well-known problem in computer science and mathematics, particularly in the theory of computation. It revolves around determining whether a given computer program will eventually stop running (halt) or continue to execute indefinitely given a particular input.
Key Points
Definition:
The halting problem is the problem of deciding, given a description of a program and an input, whether the program will halt (finish running) or run forever when executed with that input.
Alan Turing's Contribution:
The problem was first posed and proved to be undecidable by Alan Turing in 1936. In his seminal paper, Turing introduced the concept of Turing machines, which are abstract computational devices that can simulate any computer algorithm.
Undecidability:
Turing proved that a general algorithm to solve the halting problem for all possible program-input pairs cannot exist. This means that there is no universal method or machine that can determine whether any given program will halt or run indefinitely.
The proof involves a technique known as diagonalization and a self-referential paradox similar to the liar paradox.
Implications:
The undecidability of the halting problem has profound implications for computer science. It implies that there are limits to what can be computed or decided algorithmically.
It also suggests that certain questions about the behavior of computer programs cannot be answered automatically by any possible computer program.
Proof Outline:
Assume there exists a hypothetical function
H(P, I)
that determines whether programP
halts on inputI
.Construct a program
D
that usesH
to create a paradox: it halts if and only ifH
predicts it will run forever.Show that
D
leads to a contradiction, proving thatH
cannot exist.
Example
Consider a simple example:
Program P: A program that checks if the input is an even number. If it is, the program halts. If not, the program enters an infinite loop.
Input I: Suppose the input is 3.
The Halting Problem asks whether there is an algorithm that can analyze
P
andI
and correctly determine thatP
will not halt when given the inputI
.The sense of danger must not disappear:
The way is certainly both short and steep,
However gradual it looks from here;
Look if you like, but you will have to leap.Tough-minded men get mushy in their sleep
And break the by-laws any fool can keep;
It is not the convention but the fear
That has a tendency to disappear.The worried efforts of the busy heap,
The dirt, the imprecision, and the beer
Produce a few samrt wisecracke every year;Laugh if you can, but you will have to leap.
The clothes that are considered right to wear
Will not be either sensible or cheap,
So long as we consent to live like sheep
And never mention those who disappear.Much can be said for social savior-faire,
Bu to rejoice when no one else is there
Is even harder than it is to weep;No one is watching, but you have to leap.
A solitude ten thousand fathoms deep
Sustains the bed on which we lie, my dear:
Although I love you, you will have to leap;
Our dream of safety has to disappear.-- W. H. Auden
Let’s say you side-swipe your driver-side mirror. That can’t be too bad, right? It’s not, unless your side mirror is equipped with a camera. The cost of fixing a new F-150 pickup truck mirror will set you back an eye-watering $1,600. What’s worse is that “Advanced Driver Assistance Systems” — think: automatic braking and lane-keeping assistance — often lull drivers into a false sense of security and tempt them to look at their phone instead of the road.
Fifty-four percent of Americans identify as part of the middle class, including 39% who say they are “middle class” and 15% “upper-middle class.” Another 31% consider themselves “working class” and 12% “lower class.” Just 2% of U.S. adults characterize themselves as “upper class.”
Billy Oppenheimer
a) Make Your Fortune, And Then Write Your Books
Michael Lewis was an art history major at Princeton. During his senior year, while working on a 166-page thesis paper on how the Italian sculptor Donatello took inspiration from the ancient Greeks and Romans, he discovered he loved the process of researching and writing. He said, “I remember thinking, ‘I now know what I’d like to do for a living: write books.’” When he graduated in 1982, “because I hadn’t the first clue what I should write about,” he took a job at a giant Wall Street investment bank called Salomon Brothers. When Lewis got there, Wall Street was booming with chaos, excess, and “recent Princeton graduates who knew nothing about money making small fortunes,” as Lewis put it. “I stumbled into my next senior thesis.” After a year and half of carefully observing the madness, Lewis decided he was going to quit his job and write his first book. When Lewis told his boss, his boss was worried about the sanity of a person walking away from hundreds of thousands of dollars and advised Lewis to see someone to make sure he wasn’t going insane. When he told his father, his father’s advice was to wait. “Stay at Salomon Brothers for 10 years,” he said. “Make your fortune, and then write your books.” Lewis first thought back to the feeling he felt when working on his thesis, he said, “and I wanted to feel that interest in something again.” Then, he looked around at the people 10 years older than him, “and they seemed completely stuck. Their lives had completely adapted and depended on the money, the position, the status. Had I stayed 10 years, I knew I’d get trapped too. I’d lose the desire to do the other thing. I would have forgotten the feeling.” Lewis ignored the advice and bent his path toward the gratification of his desires. It took a year and half to write his first book, and during that time, “I was aware of my unknown future, but I never felt like, ‘if this thing doesn’t succeed, I’m screwed.’ It was, ‘I’m doing exactly what I want to do, and I’ll figure out a way to make it work.’” That first book was Liar’s Poker—it sold millions of copies and bent Lewis’ unknown future to conform to his desire for a life spent writing books (Moneyball, The Blind Side, The Big Short, and many others in the years since).
b) Make A Life
“Beware of looking for goals,” as Hunter S. Thompson wraps up his advice to his friend. “Look for a way of life. Determine how you want to live and then see what you can do to make a living within that way of life.” Determine what your true desires and abilities are, what captivates you, what you could be curious about learning forever, what the person that you are loves to do—then see what you can do to make a life that bends and conforms to those things.
One 2023 estimate, released by the industry group Insurance Information Institute, concluded that 12 percent of homeowners had no insurance in 2022, up from just 5 percent in 2019. Among those who own their home outright, the CFA estimates roughly 14 percent are uninsured, with low-income and minority homeowners especially at risk. Among mortgage holders, only 2 percent opt to go without coverage.
WSJ
For instance, people get more motivated for tasks when you turn the jobs into games and let them share their achievements on leaderboards. (Think of the popularity of Wordle.) It feels good to have a streak and see how you stack up to others. We might be able to transfer those competitive elements to nudged choices: If you nudge people into saving for retirement, for instance, you could show them how their savings stack up against other people’s each week.
When to use an LLM: ask them questions that are hard to answer but easy to verify. They shine in being able to pull together a lot of relevant content from many sources and synthesize it into easily-digestable form, but they have no meaningful "understanding" of what they're telling you. So they're fantastic for questions like, "Plan a 2-day vacation in London for somebody who likes museums and British history" or "Give me 20 ideas for Father's Day gifts" where you can read the result and take out the parts you like, and if it's not perfect it still saved you a lot of work and maybe came up with good ideas you wouldn't have thought of. By contrast, they would be more dangerous for "Recommend the 5 best stocks to buy for someone who's 55 and doesn't know a lot about finance," or God forbid something like, "Does drug X have interactions with drug Y?" where it may be hard or impossible to verify the accuracy of the answer and it could have grave consequences if it was wrong.
The bottom line is that for a popular service that relies on generative AI, the costs of running it far exceed the already eye-watering cost of training it. That’s because AI has to think anew every single time something is asked of it, and the resources that AI uses when it generates an answer are far larger than what it takes to, say, return a conventional search result.