Kurzweil on the singularity

Science versus philosophy in the singularity

Ray Kurzweil

Journal of Consciousness Studies, vol. 19, Nos. 7-8, (2012)

This is a somewhat rambling paper. It comprises part of a Journal of Consciousness volume devoted to the singularity. This is the point at which artificial intelligence machines already above the level of human intelligence start to produce an indefinite increase in intelligence, by means of constructing ever more intelligent computers. However, Kurzweil flits from topic to topic within consciousness studies, before finally moving to the main area of discussion in the closing section of his paper.

The author’s thoughts on some of these topics are open to question. Emotional and moral intelligence are labelled as performances that are part of the ‘easy’ as ‘opposed’ to the ‘hard’ problem of consciousness. However, this takes no account of recent neuroscientific studies showing the involvement of subjectivity in the evaluations that underlie such emotional and moral judgements.

Kurzweil goes on to deliver a lengthy diatribe against the Penrose/Hameroff quantum theory of consciousness. He is hardly alone in decrying this theory, but his argumemts could do to take a bit more account of the research of the last decade. Like others he quotes the Tegmark (2000) paper seemingly unaware that while Tegmarks calculations have not been falsified, the popularised conclusion that functional quantum states could not exist in organic matter has been falsified. This arises from discoveries from 2007 onwards of such processing in photosynthetic organisms, including multicellular organisms and organisms at room temperature. To be fair, these involve very much shorter timescales than those required by Hameroff’s version of quantum consciousness theory. Kurzweil makes the further annoying error of repeating Chalmer’s sarcastic comment that quantum consciousness theory was based on putting two mysterious things (consciousness and quantum theory) together, but claiming that this was advanced as a serious argument for the theory by Hameroff.

Kurzweil is more interesting in discussing the thought experiment as to what happens to either consciousness or the ‘self’ in a futuristic situation in which the neurons of a brain are progressively replaced by computer switches. While neurons are mainly stable, the component proteins are continually replaced so that something like the futuristic is continually happening in all our brains without any loss of continuity. The important thing appears not to be the component proteins but maintenance of the same pattern or organisation. Nevertheless, the greater stability of neurons as distinct from the rest of the body is interesting in this respect. Neurons that are selective for a particular range of images are active in conjunction with ‘hot spots’ of other neurons around them and also with global gamma synchronies.

Only in the last part of his paper does Kurzweil discuss the question of the singularity. He does not address the important question of whether either humans or computers can build systems that are more intelligent than themselves, as opposed to merely running at faster speeds. In fact Kurzweil may not even be looking for a higher level of intelligence, as in this paper he seems to be only considering the idea of existing human level intelligence plus the information storage and speed of retrieval already achieved by the much-hyped Watson computer. But it is not clear that this amounts to anything more than the current scope for human intelligence to work in conjunction with Watson-type computers. In particular there seems to be no requirement for the Watson side of the partnership to achieve consciousness or the related flexibility of preferences in organic brains.

Posted by

Leave a Reply