Miro Dyer

Someone is shy

Miro hasn't completed a profile. Should we look for some other people?

Comments & conversations

Noface
Miro Dyer
Posted about 2 years ago
Can AI help in solving the problem it's creating?
I know this might sound weird, but I think my research will bring forward significantly the singularity. Allow me to qualify my statement. I have solved the fundamental problems of AI, by doing the following- (1) believing that cognition IS computation - first that the mind is software, the brain is hardware, language is the UI (2) taking each of these areas in turn, and developing a realistic model of each. I have just finished the language solution, having completed the hardware solution last year, and the (system) software the year before that. Ten years ago, I found I had to stay at home to mind kids while my partner, a successful lawyer, pursued her (our) financial security. Therefore, if I was going to discover something basic, real and hopefully amazing, I was never going to have a better opportunity. My main quality that counts is my lateral thinking ability. I have the unconventional mindset AND the intellectual energy to pursue alternative models of cognition. I found the biggest problems to progress in this field were unnecessary - introduced by so-called gurus of the field, like Fodor. His analysis of symbol computation systems is fundamentally flawed. I have read many crazy things on the Internet that sound pretty much as I do now. I also know that a proportion of readers will always regard transhumanist pronouncements like this one with a jaundiced eye. All I ask is that you read my website, and suspend judgement until you have read and understood my work, and its main ideas. Www.tde-r.webs.com
Noface
Miro Dyer
Posted about 2 years ago
Can AI help in solving the problem it's creating?
When I went through Engineering school, as an undergraduate, our mech design professor told us that Engineers were meant to do the following algorithm- (1) innovate your job away (2) repeat step 1. That is, increases in efficiency are usually equivalent to the removal of jobs. A little thought will reveal the underlying truth of the situation. Objectively, we all benefit from the boost to the economy, but subjectively, as a father with a family, we want secure employment to carry on as long as possible. I think this is a really interesting problem in system dynamics. There will be new workers entering the workforce, as people graduate from secondary and tertiary education. But these workers are not just conservative influences who want the nipple to yield milk as long as possible- they are also young people whose tastes as a consumer change, and who apply change pressures by virtue of their mass effect on markets. There are significant infrastructural 'leaks' which add significant loads to this basic churn dynamic, the largest of which is international tax evasion systems. The large banks are the prime vectors of this loss. As always, what matters is not what is, but what can be changed. Public opinion en masse can be affected by government advertising, such as the recent non-smoking drives. Public 'background' behaviours are highly resistant to deliberate influence, yet regularly exhibit cascades, ie rapid changes in equilibrium state. This problem is chaotic. Economists sure have their work cut out. I'm an AI researcher who has made considerable inroads into the 'rusted on' problems in my field. I need my counterparts in the field of neo-economics to perform similar magic. Good luck!
Noface
Miro Dyer
Posted over 2 years ago
Robots Work, Human Beings Perform
At the level of the basic unit, the fetch-execute cycle, computers (robots have brains which are computers) are open-loop devices. The other term which is similar to open-loop is 'feedforward'. They just do stuff, and dont worry about the consequences. Humans and animals use the closed-loop principle at the basic unit level. Some people (Clark Hull, Nikolaas Tinbergen) call this 'drive-based', but the term that is used most these days is 'feedback'. These basic units do stuff too, but they are concerned about the consequences - they sense the effects of their actions, which are 'fed back', usually in order to self-regulate these actions, ie maintain homeostasis. Sometimes they are used for 'positive feedback' - a cascade, eg a sneeze, an orgasm, cell mitosis/meiosis, neuron action impulse. These two systems are not dissimilar- in fact, a feedback cycle is just a feedforward link (eg an amplifier, a neural network -really any function block) with a feedback circuit added on. Computers do have feedback circuits, but not at the hardware data level. We program them in at the software level - sometimes. As in other areas of mathematics, we have discrete (integral) types and analog (continuous) types. a discrete type of feedback is the IF-THEN loop. The IF condition is a switch- it uses fresh 'outside' (its own level) information to select which 'inside' procedural path to take. Therefore, using feedback produces computing units that are self-regulating, but also self-indeterminate. They rely on information in other units to achieve completion. However, the basic criticism of the 'von Neumann bottleneck' remains- every bit of data must pass through a narrow feedforward funnel; very, very, very fast. To use feedback to regulate and correct these data, eg at the software level, involves introducing indeterminacy, and slows the whole machine down. Turns and roundabouts. To make an intelligent computer means mastering these two systems -feedforward vs feedback.