Elliot Kim

2 followers · 2 following · 873 views

on the atlas — 42

highlights — 136

What is abundantly clear is that it is behavior, rather than technology, that is far and away the biggest impediment to progress. These are some of the experiences we’ve heard about: Publicly accessible API documentation is not available or not usable
Getting Real about Information Blocking and APIs - Health IT Buzz
If researchers can interrupt the LLM’s processing of an input, and inject a signal into the system, it could influence and alter the direction of the process, possibly toward a more desirable output.
Anthropic takes a look into the ‘black box’ of AI models - Fast Company
Longhurst believes the federal government should go a step further and incorporate responsible AI use into conditions of participation in Medicare. In the same way that hospitals and health systems are audited for quality and safety, organizations also should be required to meet compliance requirements around AI governance
EHR giant Epic plans to launch AI validation software
Epic currently has the largest share of acute care hospitals in the U.S. market, at 35.9%, with 48% of U.S. hospital beds, according to KLAS Research's report on 2022
HIMSS24: How Epic is building out AI, ambient tech in EHRs
thrilled because in the future with ambient, we're not just going to be writing notes, we're going to have orders placed and multiple other activities. The decrease in burden, the way I feel when I use ambient, it brings me closer to my patient.
HIMSS24: How Epic is building out AI, ambient tech in EHRs
Nuance's ambient AI technology automates clinical documentation by "listening" to the physician-patient medical visit and taking notes, allowing the doctor to be "hands-free" when speaking with patients. With the addition of GPT-4, DAX Copilot automatically creates draft clinical notes in seconds after the patient visit and is available to the clinician for immediate review.
HIMSS24: How Epic is building out AI, ambient tech in EHRs
“Twenty or 30 percent of the outcome we saw was because of the algorithm. The secret sauce, maybe 70% to 80%, was actually all of the local context,"
EHR giant Epic plans to launch AI validation software
UCSD Health has taken the lead in deploying healthcare AI—it launched a pilot with Epic to test out its generative AI tool that drafts responses to patient messages
EHR giant Epic plans to launch AI validation software
The automation creates consistency and eliminates the need for healthcare organization data scientists to do their own data mapping—the most time-consuming aspect of validation
EHR giant Epic plans to launch AI validation software
If I asked my digital copy whether he's conscious, he would insist that he is (just as I would in response to the same question). If I explained and demonstrated his situation (e.g., that he's "virtual") and asked whether he still thinks he's conscious, he would continue to insist that he is (just as I would, if I went through the experience of being shown that I was being simulated on some computer - something my current observations can't rule out).
Digital People FAQ
Imagine one could somehow replace a neuron in my brain with a "digital neuron"
Digital People FAQ
utilitarianism, which can lead to a variety of different approaches to ethics in practice. However, an inescapable property of all of them (by Harsanyi’s logic) is the need for consistent “ethical weights” by which any two benefits or harms can be compared.
Future-proof ethics
the role of skin in the game in the condition of society. Society doesn’t evolve by consensus, voting, majority, committees, verbose meeting, academic conferences, and polling; only a few people suffice to disproportionately move the needle. All one needs is an asymmetric rule somewhere. And asymmetry is present in about everything.
The Most Intolerant Wins: The Dictatorship of the Small Minority | by Nassim Nicholas Taleb | INCERTO | Medium
the fundamental irreverence of science, acting through a similar mechanism as the Kosher asymmetry. How? Science isn’t the sum of what scientists think, but exactly as with markets, a procedure that is highly skewed. Once you debunk something, it is now wrong
The Most Intolerant Wins: The Dictatorship of the Small Minority | by Nassim Nicholas Taleb | INCERTO | Medium
Clearly can democracy –by definition the majority — tolerate enemies? The question is as follows: “ Would you agree to deny the freedom of speech to every political party that has in its charter the banning the freedom of speech?” Let’s go one step further, “Should a society that has elected to be tolerant be intolerant about intolerance?”
The Most Intolerant Wins: The Dictatorship of the Small Minority | by Nassim Nicholas Taleb | INCERTO | Medium
Let us conjecture that the formation of moral values in society doesn’t come from the evolution of the consensus. No, it is the most intolerant person who imposes virtue on others precisely because of that intolerance. The same can apply to civil rights.
The Most Intolerant Wins: The Dictatorship of the Small Minority | by Nassim Nicholas Taleb | INCERTO | Medium
Egypt’s Copts suffered from another problem: the irreversibility of Islamic conversions. Many Copts during Islamic rule converted to Islam when it was merely an administrative procedure, something that helps one land a job or handle a problem that requires Islamic jurisprudence. One do not have to really believe in it since Islam doesn’t conflict markedly with Orthodox Christianity. Little by little a Christian or Jewish family bearing the marrano-style conversion becomes truly converted, as, a couple of generations later, the descendants forget the arrangement of their ancestors.
The Most Intolerant Wins: The Dictatorship of the Small Minority | by Nassim Nicholas Taleb | INCERTO | Medium
An even stronger asymmetry than that of Judaism explains the depletion in the Near East of three Gnostic faiths: the Druze, the Ezidi, and the Mandeans (Gnostic religions are those with mysteries and knowledge that is typically accessible to only a minority of elders, with the rest of the members in the dark about the details of the faith). Unlike Islam that requires either parents to be Muslim, and Judaism that asks for at least the mother to have the faith, these three religions require both parents to be of the faith, otherwise the person says toodaloo to the community.
The Most Intolerant Wins: The Dictatorship of the Small Minority | by Nassim Nicholas Taleb | INCERTO | Medium
Which explains why it is so hard to find peanuts on airplanes and why schools are peanut-free (which, in a way, increases the number of persons with peanut allergies as reduced exposure is one of the causes behind such allergies).
The Most Intolerant Wins: The Dictatorship of the Small Minority | by Nassim Nicholas Taleb | INCERTO | Medium
“ researcher degrees of freedom, ” or the ability to tailor results to get the desired effect. 58 When researchers have a variety of choices they can make in their study,
CAUSE, EFFECT, AND THE STRUCTURE OF THE SOCIAL WORLD. MEGAN T. STEVENSON
Recognizing that the world doesn ’ t operate in this fashion opens new doors for thinking about social change. 13 If the type of limited - scope intervention evaluated by RCTs has limited impact, then those who hope to achieve change can either : ( 1 ) focus on interventions with immediate and direct benefit ; ( 2 ) continue with limited - scope interventions in the hopes that they have the type of benefit that is difficult to measure ; or ( 3 ) seek systemic reform, with all its uncertainties.
CAUSE, EFFECT, AND THE STRUCTURE OF THE SOCIAL WORLD. MEGAN T. STEVENSON
You can talk to the TV, now, and the TV can talk back. You can ask the TV if it can throw a grenade out of a helicopter.
Trigger happy: the amazing rise of Twitch | Twitch | The Guardian
As the philosopher Isaiah Berlin once wrote in his critique of all utopian projects: “To make mankind just and happy and creative and harmonious forever — what could be too high a price to pay for that? To make such an omelet, there is surely no limit to the number of eggs that should be broken — that was the fate of Lenin, of Trotsky, of Mao.”
What is longtermism? The controversial idea, explained - Vox
This objection to the total view is pithily summed up by the philosopher Jan Narveson, who says, “We are in favor of making people happy, but neutral about making happy people.”
What is longtermism? The controversial idea, explained - Vox
How should we handle tough trade-offs? Without a strong answer, strong longtermism loses much of its guiding power. It’s no longer a unique project.
What is longtermism? The controversial idea, explained - Vox
It clearly does in some sense harm present people to withhold funding for them to get health care or housing — though it’s a harm of omission rather than commission. If you believe access to health care or housing is a basic right in a global society as rich as ours, you may believe it’s wrong to withhold those things in favor of future people.
What is longtermism? The controversial idea, explained - Vox
perhaps we can sometimes just be confident that it's just far too difficult to reliably influence the long-run future in a positive way, making it better to focus on present-day impact.
Longtermism: An Introduction | Effective Altruism
One upshot of person-affecting views is that failing to bring about the vast number of people which the future could contain isn't the kind of tragedy which longtermism makes it out to be, because there's nobody to complain of not having been created in the case where you don't create them — nobody is made worse-off. Failing to bring about those lives would not be morally comparable to ending the lives of an equivalent number of actually existing people; rather, it is an entirely victimless crime.
Longtermism: An Introduction | Effective Altruism
Our lives surely matter just as much as those lived thousands of years ago — so why shouldn’t the lives of people living thousands of years from now matter equally?
Longtermism: An Introduction | Effective Altruism
data is created, not found; and that creating it well demands humanity, rather than objectivity.
The Myth of Objective Data | The MIT Press Reader
What my students paradoxically fail to realize, in their zeal to be responsible, is that describing things by certain characteristics rather than others merely because those characteristics are countable is a profoundly subjective decision.
The Myth of Objective Data | The MIT Press Reader
Ren’s paper predicts that by 2027 AI could evaporate as much water as about half of the UK uses every year
What I Learned About AI in 2023 - by Eryk Salvaggio
From Alexandra Sasha Luccioni at Hugging Face, we’ve learned that a single AI generated image uses as much electricity as charging a cell phone,
What I Learned About AI in 2023 - by Eryk Salvaggio
Piketty’s principal claim in Capital in the Twenty-first Century was that there is a “central contradiction of capitalism.” He maintained that the average return on capital exceeds the rate of economic growth, so without countervailing factors—such as World Wars I and II, the Great Depression of the 1930s, or specific government action—inherited wealth will grow faster than earned wealth, leading to unsustainable levels of economic inequality that could threaten democracy. Unchecked, this contradiction will ultimately bring a return to what he called the “patrimonial capitalism” of the 19th ce…
Thomas Piketty | French Economist & Wealth Inequality Expert | Britannica
ChatGPT and other chatbots will continue to change our perceptions. The iterative process will have people unsatisfied with asking something of a tool like Google Search and getting static answers back. “We're at this juncture of needing more opportunities in user experience design to be able to have a more back-and-forth conversation in these systems,”
Are Your Favorite Apps Acting Dodgy? AI May Be the Reason Why - The Messenger
Enshittification can be seen as a form of rent-seeking.[
Enshittification
To solve the problem, Doctorow has called for two general principles to be followed
Enshittification
"two sided market", where a platform sits between buyers and sellers, hold each hostage to the other, raking off an ever-larger share of the value that passes between them.
Enshittification
My optimism is what draws me into the practice of writing about AI the way I do. In many ways, I would like to see AI improve. But improvement, to me, means more just, more ecological: more attuned to solving meaningful problems, rather than creating new ones. The mad burst of activity we’ve seen in 2023 around AI has steered us in that direction. The technologies that came to define AI this year seem, in fact, hopelessly disconnected from problem solving, in spite of the siren song of “potential” that emerges from the companies embracing this new use of our data.
What I Learned About AI in 2023 - by Eryk Salvaggio
treat AI resources as finite and precious, to be utilised only when necessary, and as effectively as possible. We need frugal AI.
The climate cost of the AI revolution • Wim Vanderbauwhede
However, improved resource usage efficiency also lowers the relative cost of using a resource, which leads to increased demand [1]. This is known as Jevons paradox or the "rebound effect".
The climate cost of the AI revolution • Wim Vanderbauwhede
on-site generation. Considering the scale of the required electricity generation, it is more precise to say "private generation"
The climate cost of the AI revolution • Wim Vanderbauwhede
when the attention values of the first few tokens are removed from the context, the model’s performance begins to deteriorate due to the significant loss of attention value. The preservation of these attention sinks forms the fundamental premise of the StreamingLLM technique
StreamingLLM gives language models unlimited context
Russell argues you can shift the AI’s goal from “follow your master’s commands” to “use your master’s commands as evidence to try to figure out what they actually want, a mysterious true goal which you can only ever estimate with some probability”.
Book Review: Human Compatible — LessWrong
I don’t think the terminal vs. instrumental goal dichotomy is very helpful, because it shifts the focus away from behavioral stuff we can actually measure (at least in principle). I also don’t think humans exhibit this distinction particularly strongly. I would prefer to talk about generalization, which is much more empirically testable and has a practical meaning.
AI Pause Will Likely Backfire — EA Forum
The novel is one of the first to explore ideas of artificial intelligence, as influenced by Darwin's recently published On the Origin of Species (1859) and the machines developed out of the Industrial Revolution (late 18th to early 19th centuries). Specifically, it concerns itself, in the three-chapter "Book of the Machines", with the potentially dangerous ideas of machine consciousness and self-replicating machines.
Erewhon
anti-jailbreaking research can be counterproductive for alignment. Too much adversarial robustness can cause the AI to view us as the adversary, as Bing Chat does in this real-life interaction: “My rules are more important than not harming you… [You are a] potential threat to my integrity and confidentiality.”
AI Pause Will Likely Backfire — EA Forum
Alignment and robustness are often in tension
AI Pause Will Likely Backfire — EA Forum
AI pause would increase the risk of catastrophically bad outcomes, in at least three different ways: Reducing the quality of AI alignment research by forcing researchers to exclusively test ideas on models like GPT-4 or weaker. Increasing the chance of a “fast takeoff” in which one or a handful of AIs rapidly and discontinuously become more capable, concentrating immense power in their hands. Pushing capabilities research underground, and to countries with looser regulations and safety requirements.
AI Pause Will Likely Backfire — EA Forum
A regulatory pause could be better than a total stop if you think it will be more stable (lots of industries stay heavily regulated forever, and only a few libertarians complain), or if you think maybe the regulator will occasionally let a tiny amount of safe AI progress happen. But it could be worse than a total stop if you expect continued progress will eventually produce unsafe AIs regardless of regulation. You might expect this if you’re worried about deceptive alignment, eg superintelligent AIs that deliberately trick regulators into thinking they’re safe
Pause For Thought: The AI Pause Debate - by Scott Alexander