Activity: Talk or presentation › Oral presentation
With the rise of A.I., expert-systems, machine-learning technology and Big Data we may start to wonder whether people as creative, cognitive and intellectual beings will become redundant for the generation of knowledge. Also, the increasing success of machine-learning technology in finding patterns in data makes us ask whether scientific theories will become superfluous. First, it will be argued that an empiricist conception of science makes these thoughts plausible and provides little room to criticize them. Next, Michael Polanyi’s (1958/1962) notion of personal knowledge will be taken as a starting-point in the development of a suitable alternative that avoids the rather defensive position in which an empiricist is forced and preferably does not fall back on naive scientific realism. Hence, this paper aims at a contribution to critically investigating whether theoretical knowledge and the scientist’s role in developing it, will remain crucial – or will arbitrary algorithms, provided by machine-learning technologies for constructing relationships between data-input-output, eventually be able to meet crucial epistemic criteria such as empirical adequacy, reliability and relevance, better than limited humans ever could? The empiricist strand in the philosophy of science has a long history of making the role of humans in science superfluous, or at least, to downplay their role such as to justify the objectivity of knowledge. Strict empiricism, from Hume to Logical Positivism and anti-realist views such as Van Fraassen, can only assume that, as theory-formation transcends what has been given in empirical data, theories must be understood as just heuristic tools that are only necessary for limited beings. We may also recall the responses to Bogen and Woodward (1988) when claiming that phenomena can be found in data. McAllister (1997) criticizes this view by arguing that there are always infinitely many patterns in any data-set, and so the choice of one as being a phenomenon is subjectively stipulated by the investigator. Glymour's (2000) strict empiricist alternative is that scientists infer from data to patterns by means of statistical analysis, which, according to him, does not involve subjective grounds, and which does not add anything new to the data. Clearly, if, objectively, there is just patterns in data-sets, only subjectively summarized in theories (including ‘mini-theories’ such as phenomena and laws), it is to be expected that machines will surpass the intellectual contributions of humans. Polanyi was a physical-chemist, who proposed a philosophical conception of science that accounted for the role of the scientist. Central is his premise that the supposed objective-subjective gap has led to the flawed belief – in positivism and empiricism – that experiences are the objectively reliable part of knowledge, instead of theory. Polanyi defends personal knowledge, in which the contribution of the human intellect to knowledge is crucial. He bridges the objective-subjective gap by claiming that science must be understood as an inherently and inescapably responsible endeavor, because individual scientists necessarily subject themselves to general rules and ‘out of passion’ take responsibility for convincing others of their theoretical findings. This paper aims to explain the relevance of Polanyi’s argument for better understanding the indispensable role of the human intellect.