Hi Alan In my view, what individual researchers, such as ourselves, are doing is of vital importance to the development of future, commercial, AGI applications. We are the post-modernist skunkworx it seems, the garage Apple of tomorrow working with the soul of the first, machine species. Homo Mechanicus?
True, we are not part of the commercial AGI wave now. It is highly unlikely we would be funded for our efforts. However, in my version of a working AGI solution, the 'HOW TO' of a physical architecture would definitely come from specialized sources, such as yourself. My contention has been all along that the progress of anything AGI has been stymied by the lack of a workable, architectural blueprint. To me, that is the general gap of the day, which must be filled. Researchers need a framework to slot their R&D into, not so? That is my area of special research, to quantify concepts such as intelligence, consciousness, machine reality, and so on, all within the context of AGI. Many here have their respective, specialized areas. Once a workable envelope has been created, we would be able to connect these, and other, dots to the physical operation of such a platform. The tapestry would yet emerge. Keep the faith. Please carry on with your passion and research. We all need to. I'm totally convinced that all our work would ultimately find a place in the AGI (his)story. In my view, the sooner, the better. None of us are getting any younger. Robert ________________________________ From: Alan Grimes <[email protected]> Sent: Thursday, 18 October 2018 6:30 AM To: Matt Mahoney Subject: [agi] Too soon to start thinking like an early adopter? While it doesn't look like there is any possibilty for me to get in on the development of AGI beyond mouthing off on this list, getting in as early as I can on the adoption side seems to be about the least stupid thing I can do, if I can do it right... Right now the basic model of AI are cloud services where the usual suspects get you to buy a listening device, place it in your home, and basically use the microphone to spy on your life. With AGI, you really want your own rig, especially when using it with a neural interface. In that case, you would want a cloud service as a redundant fail-over, but you would still want your own equipment for normal use. While it's almost certain that hardware will evolve FAST to the point that only nano-tech re-building the equipment every ~40 mintues or so could possibly keep up... In terms of technology, it's certain to be an S-curve that will reach some optimum configuration of quarks and gluons (barring some radical discovery about the nature of the universe)... In terms of scale, well I hope it leaves room for humanoids and or other types of life including biological life and quazi-biological life. In case you haven't noticed, it's late 2018 already. The Big Weenie says AI is coming in 2029. What I'm seeing from Deep Mind is already vastly superhuman but only lacks the high level organization required to be generally intelligent. So once we reach the point of "takeoff" it's going to be a pretty steep coming right out of the gates. =O The thing is that the singularity is now within the planning window of my next computer upgrade, which I typically use for 5-7 years. =P So I am thinking of getting a threadripper R9X system which has a massive number of PCI-E lanes under the theory that the first few iterations of AGI hardware will be PCI-E compatible, which will lead to a rack server which I can tie in from the basement which will be followed by ??? So is my thinking off base or should I move ahead with this plan? -- Please report bounces from this address to [email protected] Powers are not rights. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Tacbd56e2434d9021-Mf61b594515a6e3376d614e5e Delivery options: https://agi.topicbox.com/groups/agi/subscription
