= 27d567bc2d48d9f40e9ccc20ea370b15)
My understanding is that CEA exists in order to simplify the paperwork of multiple projects. For example, Effective Animal Activism is not its own charity; instead, you donate to CEA and transfer the money to EAA. As bryjnar said, there's not really any overhead in doing this. Using CEA as an umbrella much simpler than trying to get 501(c)(3) status for EAA on its own, which would be painstaking process.
I am disappointed that my realistic and fact based observation generated a down vote.
At the risk of an additional down vote, but in the interest of transparent honest exchange, I am pointing out a verifiable fact, however unsavory it may be interpreted.
If over time the time cost of intermediaries (additional handling and overhead costs) remains below the cost of the steps to eliminate intermediaries (the investment required to establish a 501(c)(3)) then I stand corrected. While an improbable situation, it could well be possible.
In response to
How minimal is our intelligence?
200k years ago when Homo Sapiens first appeared, fundamental adaptability was the dominant force. The most adaptable, not the most intelligent, survived. While adaptability is a component of intelligence, intelligence is not a component of adaptability. The coincidence with the start of the ice age is consistent with this. The ice age is a relatively minor extinction event, but none the less the appearance and survival of Homo Sapiens is consistent, where less adaptable life forms did not survive.
Across the Hominidae family Homo Sapiens proved to be most adaptable. During the ice age the likely focus was simply to survive. When a temperate climate returned there are some who believe Homo Sapiens, much as future Aztecs and others, began to systematically eliminate their competition.
Concurrently, another phenomenon was occurring. Homo Sapiens was learning and steadily increasing their understanding of the world. While there is not evidence that has survived the years, it would be reasonable to posit that learning continued in much the same fashion as today; new knowledge building on established knowledge. Being less organized than later situations it would progress more slowly.
Our improved knowledge likely increased our survival rates through the second ice age. When temperate climates returned, the stage was set for the advancement of mankind to organized farming, written language and Ur.
Somewhere in this time frame, intelligence began to overtake adaptability as the dominant force. This also marked the shift from evolutionary pressure to societal pressure as the underlying force behind advancement and survivability. The random nature of evolutionary advances gave way to a more complex society-driven selection process.
It's also important to draw a subtle distinction. The advances were not a function of increase in general IQ. They were a function of integration of the concepts envisioned by a subset of high IQ individuals into society; i.e., a societal variant of evolutionary adaptability.
In response to
Logical Pinpointing
I'm new here, so watch your toes...
As has been mentioned or alluded to, the underlying premise may well be flawed. By considerable extrapolation, I infer that the unstated intent is to find a reliable method for comprehending mathematics, starting with natural numbers, such that an algorithm can be created that consistently arrives at the most rational answer, or set of answers, to any problem.
Everyone reading this has had more than a little training in mathematics. Permit me to digress to ensure everyone recalls a few facts that may not be sufficiently appreciated. Our general education is the only substantive difference between Homo Sapiens today and Homo Sapiens 200,000 years ago.
With each generation the early education of our offspring includes increasingly sophisticated concepts. These are internalized as reliable, even if the underlying reasons have been treated very lightly or not at all. Our ability to use and record abstract symbols appeared at about the same time as farming. The concept that "1" stood for a single object and "2" represented the concept of two objects was establish along with a host of other conceptual constructs. Through the ensuing millennia we now have an advanced symbology that enables us to contemplate very complex problems.
The digression is to point out that very complex concepts, such as human logic, require a complex symbology. I struggle with understanding how contemplating a simple artificially constrained problem about natural numbers helps me to understand how to think rationally or advance the state of the art. The example and human rationality are two very different classes of problem. Hopefully someone can enlighten me.
There are some very interesting base alternatives that seem to me to be better suited to a discussion of human rationality. Examining the shape of the Pareto front generated by PIBEA (Prospect Indicator Based Evolutionary Algorithm for Multiobjective Optimization Problems) runs for various real-world variables would facilitate discussions around how each of us weights each variable and what conditional variables change the weight (e.g., urgency).
Again, I intend no offense. I am seeking understanding. Bear in mind that my background is in application of advanced algorithms in real-world situations.
In response to
[Proposed Paper] Predicting Machine Super Intelligence
There are some good ideas in this.
The paper needs focus. One possibility is the technique described in the abstract "The concept of MSI is dissected..... a model is constructed..." Is there a specific formal technique that you are going to use?
Another possibility is a review of prediction techniques, with an attempt to apply each one to full AI, or references that do so. Sotala and Armstrong surveyed predicted dates to AI; you could survey the different techniques one could use or which have been used.
It seems that the section of the abstract that analyzes accelerated change ("Rate modifiers") could be omitted as off-topic to either of the two possibilities above. Given what appears to be the main topic, I would suggest keeping the review of the AI risk short; and not going into too much detail into specific technologies like AIXI or the Goedel machine. I am not too sure about the componentry section, given that we have no idea what components might be needed.
Joshua,
Thank you for the feedback.
I do need to increase the emphasis on the focus, which is the first premise you mentioned. I did not do that in this draft with the intent of eliciting feedback on the viability and interest in the model concept.
I will use formal techniques, which one(s) I have not yet settled on. At the moment, I am leaning to the processes around use case development to decompose current AI models into the componentry. For the weighting and gap calculations some statistical methods should help.
I am mulling over Bill Hibbard's 2012 AGI papers, "Avoiding Unintended AI Behaviors" and "Decision Support for Safe AI Design" http://www.ssec.wisc.edu/~billh/g/mi.html as well as some PIBEA findings, e.g., http://www.cs.umb.edu/~jxs/pub/cec11-prospect.pdf to use as a framework for the component model. The Pareto front element is particularly interesting when considered with graph theory.
I am considering how the rate modifiers can be incorporated into the predictive model. This will help to identify what events for the community to look for and how a rate modifier occurrence in one area, e.g., pattern recognition, impacts other aspects of the model. We clearly do not know all of the components, but we do know the major disciplines that will contribute. As noted, the model will be extensible to allow discoveries to be incorporated, increasing the accuracy.
The general idea is to establish a predictive model with assumed margins of error and functionality. To put a formalized "stick in the ground" from which improvements are made. If maintained and enhanced with discoveries the margin of error will continue to decline and confidence levels will increase. Such a model also provides context for research and identifies potential areas of study.
One potential aspect of the model is to identify aspects of research that may be obviated. If a requirement is consistently satisfied through unexpected methods, it can be removed from consideration in the area where it was originally conceived. This also has the potential to provide insights to the original space.
[Proposed Paper] Predicting Machine Super Intelligence
Note from Malo
The Singularity Institute is always on the lookout for interested and passionate individuals to contribute to our research. As Luke frequently reminds everyone, we've got 2–3 years of papers waiting to be written (see “Forthcoming and Desired Articles on AI Risk”). If you are interested in contributing, I want to hear from you! Get in touch with me at malo@intelligence.org
We wish we could work with everyone who expresses an interest in contributing, but that isn't feasible. To provide a path to becoming a contributor we encouraging individuals to read up on the field, identify an article they think they could work on, and post a ~1000 word outline/preview to the LW community for feedback. If the community reacts positively (based on karma and comments) we'll support the potential contributors' effort to complete the paper and—if all goes well—move forward with an official research relationship (e.g.,Visiting Fellow, Research Fellow or Research Associate).
Hello,
This is my first posting here, so please forgive me if I make any missteps.
The outline draft below draws heavily on Intelligence Explosion: Evidence and Import (Muehlhauser and Salamon 2011?). I will review Stuart Armstrong’s How We're Predicting AI... or Failing to, (Armstrong 2012) for additional content and research areas.
I'm not familiar with the tone and tenor of this community, so I want to be clear about feedback. This is an early draft and as such, nearly all of the content may or may not survive future edits. All constructive feedback is welcome. Subjective opinion is interesting, but unlikely to have an impact unless it opens lines of thought not previously considered.
I'm looking forward to a potentially lively exchange.
Jay
Predicting Machine Super Intelligence
Jacque Swartz
Most Certainly Not Affiliated with Singularity Institute
jaywswartz@gmail.com
Abstract
This paper examines the disciplines, domains, and dimensional aspects of Machine Super Intelligence (MSI) and considers multiple techniques that have the potential to predict the appearance of MSI. Factors that can impact the speed of discovery are reviewed. Then, potential prediction techniques are considered. The concept of MSI is dissected into the currently comprehended components. Then those components are evaluated to indicate their respective state of maturation and the additional behaviors required for MSI. Based on the evaluation of each component, a gap analysis is conducted. The analyses are then assembled in an approximate order of difficulty, based on our current understanding of the complexity of each component. Using this ordering, a collection of indicators is constructed to identify an approximate progression of discoveries that ultimately yield MSI. Finally, a model is constructed that can be updated over time to constantly increase the accuracy of the predicted events, followed by conclusions.
I. Introduction
Predicting the emergence of MSI could potentially be the most important pursuit of humanity. The distinct possibility of an MSI emerging that could harm or exterminate the human race (citation) demands that we create an early warning system. This will give us the opportunity to ensure that the MSI that emerges continues to advance human civilization (citation).
We currently appear to be at some temporal distance from witnessing the creation of MSI (multiple citations). Many factors, such as a rapidly increasing number of research efforts (citation) and motivations for economic gain (citation), clearly indicate that there is a possibility that MSI could appear unexpectedly or even unintentionally (citation).
Some of the indicators that could be used to provide an early warning tool are defined in this paper. The model described at the end of the paper is a potentially viable framework for instrumentation. It should be refined and regularly updated until a more effective tool is created or the appearance of MSI.
This paper draws heavily upon Intelligence Explosion: Evidence and Import (Muehlhauser and Salamon 2011?) and Stuart Armstrong’s How We're Predicting AI... or Failing to, (2012).
This paper presupposes that MSI is generally understood to be equivalent to Artificial General Intelligence (AGI) that has developed the ability to function at levels substantially beyond current human abilities. The latter term will be used throughout the remainder of this paper.
II. Overview
In addition to the fundamental challenge of creating AGI, there are a multitude of theories as to the composition and functionality of a viable AGI. Section three explores the factors that can impact the speed of discovery in general. Individual indicators are explored for unique factors to consider. The factors identified in this section can radically change the pace of discovery.
The fourth section considers potential prediction techniques. Data points and other indicators are identified for each prediction model. The efficacy of the models is examined and developments that increase a model’s accuracy are discussed.
The high degree of complexity of AGI indicates the need to subdivide AGI into its component parts. In the fifth section the core components and functionality required for a potential AGI are established. Each of the components is then examined to determine its current state of development. Then an estimate of the functionality required for an AGI is created as well as recording of any identifiable dependencies. A gap analysis is then performed on the findings to quantify the discoveries required to fill the gap.
This approach does increase the likelihood of prediction error due to the conjunction fallacy, exemplified by research such as the dice selection study (Tversky and Kahneman 1983) and covered in greater detail by Eliezer Yudkowski’s bias research (Yudkowski 2008). Fortunately, the exposure to this bias diminishes as each component matures to its respective usability point and reduces the number of unknown factors.
The sixth section examines the output of the gap analyses for additional dependencies. Then the outputs are assembled in an approximate order of difficulty, based on our current understanding of the complexity of each output. Using this ordering, combined with the dependencies, a collection of indicators with weighting factors is constructed to identify an approximate progression of discoveries that ultimately yield AGI.
Comprehending the indicators, dependencies and rate factors in a model as variables provides a means, however crude, to reflect their impact when they do occur.
In the seventh section, a model is constructed to use the indicators and other inputs to estimate the occurrence of AGI. It is examined for strengths and weaknesses that can be explored to improve the model. Additional enhancements to the model are suggested for exploration.
The eighth and final section includes conclusions and considerations for future research.
III. Rate Modifiers
This section explores the factors that can impact the speed of discovery. Individual indicators are explored for unique factors to consider. While the factors identified in this section can radically change the pace of discovery, comprehending them in the model as variables provides a means to reflect their impact when they do occur.
Decelerators
Discovery Difficulty
Disinclination
Lower Probability Events
Societal Collapse
Fraud
++
Accelerators
Improved Hardware
Better Algorithms
Massive Datasets
Progress in Psychology and Neuroscience
Accelerated Science
Collaboration
Crossover
Economic Pressure
Final Sprint
Outliers
Existing Candidate Maturation
++
IV. Prediction Techniques
This section considers potential prediction techniques. Some techniques do not require the indicators above. Most will benefit by considering some or all of the indicators. It is very important to not loose sight of the fact that mankind is inclined to inaccurate probability estimates and overconfidence (Lichtenstein et al. 1992; Yates et al. 2002)
Factors Impacting Accurate Prediction
Prediction Models
Wisdom of Crowds
Hardware Extrapolation
Breakthrough Curve
Evolutionary Extrapolation
Machine Intelligence Improvement Curve
++
V. Potential AGI Componentry
This section establishes a set of core components and functionality required for a potential AGI. Each of the components is then examined to determine its current state of development as well as any identifiable dependencies. Then an estimate of the functionality required for a AGI is created followed by a gap analysis to quantify the discoveries required to fill the gap.
There are various existing AI implementations as well as AGI concepts currently being investigated. Each one brings in unique elements. The common elements across most include; decision processing, expert systems, pattern recognition and speech/writing recognition. Each of these would include discipline-specific machine learning and search/pre-processing functionality. There also needs to be a general learning function for addition of new disciplines.
Within each discipline there are collections of utility functions. They are the component technologies required to make the higher order discipline efficient and useful. Each of the elements mentioned are areas of specialized study being pursued around the world. They draw from an even larger set of specializations. Due to complexity, in most cases there are second-order, and more, specializations.
Alternative Componentry
There are areas of research that have high potential for inserting new components or substantially modifying the comprehension of the components described.
Specialized Componentry
Robotics and other elements.
Current State
Decision Processing
Expert Systems
Pattern Recognition
Speech/Writing Recognition
Machine Learning
Decision Processing
Expert Systems
Pattern Recognition
Speech/Writing Recognition
Search/Pre-Processing
Decision Processing
Expert Systems
Pattern Recognition
Speech/Writing Recognition
Target State
The behaviors required for an AGI to function with acceptable speed and accuracy are not precise. The results of this section are based on a survey of definitions from available research.
Decision Processing
Expert Systems
Pattern Recognition
Speech/Writing Recognition
Machine Learning
Decision Processing
Expert Systems
Pattern Recognition
Speech/Writing Recognition
Search/Pre-Processing
Decision Processing
Expert Systems
Pattern Recognition
Speech/Writing Recognition
Dependencies
Gap Analysis
VI. Indicators
The second section examines the output of the gap analyses for additional dependencies. Then the outputs are assembled in an approximate order of difficulty, based on our current understanding of the complexity of each output. Using this ordering, combined with the dependencies, a collection of indicators is constructed to identify an approximate progression of discoveries that ultimately yield an AGI.
Additional Dependencies
Complexity Ranking
Itemized Indicators
VII. Predictive Model
In this section, a model is constructed using the indicators and other inputs to estimate the occurrence of AGI. It is examined for strengths and weaknesses that can be explored to improve the model. Additional enhancements to the model are suggested for exploration.
The Model
Strengths & Weaknesses
Enhancements
VIII. Conclusions
Based on the data and model created above the estimated time frame for the appearance of AGI is from x to y. As noted throughout this paper, the complex nature of AGI and the large number of discoveries and events that need to be quantified using imperfect methodologies, a precise prediction of when AGI will appear is currently impossible.
The model developed in this paper does establish a quantifiable starting point for the creation of an increasingly accurate tool that can be used to continually narrow the margin of error. It also provides a starting set of indicators that can serve as early warning of AGI when discoveries and events are made.
The sheer volume of the scenarios diminishes the likelihood of any one of them. The numerous variations indicate an intractable predictability. While subject to conjunction bias, a more granular approach is the only feasible method to determine even a hint of the pre-FAI environment. Only a progressively refined model can provide information of value.
As I noted on the 80,000 Hours thread, intermediaries are nearly always an added expense on the distribution side. In this case, distribution of donations. The immediate impact is that fewer donation dollars (or whatever currency) find their way to the target organizations. The exception is if an intermediate organization facilitates a 100% pass-through, due to other funding or altruistic efforts.
In response to
Giving What We Can, 80,000 Hours, and Meta-Charity
I am compelled to point to a fundamental supply chain issue; intermediary drag. Simply stated, the greater the number of steps, the greater the overhead expense. While aggregators have some advantage on the purchasing side, they are an added expense on the distribution side in the vast majority of cases. If they enable some form of extended access, intermediaries may have a value, but the limited nature of charitable donations would make intermediaries an unlikely advantage.
In response to
Welcome to Less Wrong! (July 2012)
Hello,
I am Jay Swartz, no relation to Aaron. I have arrived here via the Singularity Institute and interactions with Louie Helm and Malo Bourgon. Look me up on Quora to read some of my posts and get some insight to my approach to the world. I live near Boulder, Colorado and have recently started a MeetUp; The Singularity Salon, so look me up if you're ever in the area.
I have an extensive background in high tech, roughly split between Software Development/IT and Marketing. In both disciplines I have spent innumerable hours researching human behavior and thought processes in order to gain insights into how to create user interfaces and how to describe technology in concise ways to help people to evaluate the merits of the technology. I've spent time at Apple, Sun, Seagate, Mensa, Osborne and a few start-ups applying my ever-deepening understanding of the human condition.
Over the years, I have watched synthetic intelligence (I much prefer the more precise SI over AI) grow in fits and starts. I am increasing my focus in this area because I believe we are on the cusp of general SI (GSI). There is a good possibility that within my life time I will witness the convergence of technology that leads to the appearance of GSI. This will in part be facilitated by advances in medicine that will extend my lifespan well past 100 years.
I am currently building my first SI web crawler that will begin building a corpus to be mined by some SciPy applications I have on my list of things to do. These efforts will provide me with technical insights on the SI challenge. There is even the possibility, however slight, that they can be matured to make a contribution to the creation of SI.
Finally, I am working on a potential paper for the Singularity Institute. I just posted a first outline/draft, Predicting Machine Super Intelligence, but do not yet know the details on how anyone finds it or how I see any responses. Having been on more than a few sites similar to this, I know I will be able to quickly sort thing out.
I am looking forward to reading and exchanging ideas here. I will strive to contribute as much as I receive.
Jay
= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
= f037147d6e6c911a85753b9abdedda8d){kind=logo}
This doesn't follow. Epsilon is by definition arbitrary, therefore I could say that I want it to be 1 / 4^^^4 if I want to.
If we accept Eliezer's proposition that the disutility of a dust speck is > 0, this doesn't prevent us from deciding that it is < epsilon when assigning a finite disutility to 50 years of torture.
For a site promoting rationality this entire thread is amazing for a variety of reasons (can you tell I'm new here?). The basic question is irrational. The decision for one situation over another is influenced by a large number of interconnected utilities.
A person, or an AI, does not come to a decision based on a single utility measure. The decision process draws on numerous utilities, many of which we do not yet know. Just a few utilities are morality, urgency, effort, acceptance, impact, area of impact and value.
Complicating all of this is the overlay of life experience that attaches a function of magnification to each utility impact decision. There are 7 billion, and growing, unique overlays in the world. These overlays can include unique personal, societal or other utilities and have dramatic impact on many of the core utilities as well.
While you can certainly assign some value to each choice, due to the above it will be a unique subjective value. The breadth of values do cluster in societal and common life experience utilities enabling some degree of segmentation. This enables generally acceptable decisions. The separation of the value spaces for many utilities preclude a single, unified decision. For example, a faith utility will have radically different value spaces for Christians and Buddhists. The optimum answer can be very different when the choices include faith utility considerations.
Also, the circular example of driving around the Bay Area is illogical from a variety of perspectives. The utility of each stop is ignored. The movement of the driver around the circle does not correlate to the premise that altruistic actions of an individual are circular.
For discussions to have utility value relative to rationality, it seems appropriate to use more advanced mathematics concepts. Examining the vagaries created when decisions include competing utility values or are near edges of utility spaces are where we will expand our thinking.