in a world where it is likely that machine learning and artificial intelligence can figure out anything it is understandable that not everything is knowable a recent study by a team of ai researchers and mathematicians suggests that even the most intelligent algorithms are bound by the constraints of mathematics inspite of the boundless potential of machine learning the power of mathematics however comes with a constraint that it is possible to prove everything the first computer scientist and author shai ben david from the university of waterloo led a research which talks about the awareness that mathematical limitations are often tied to incompleteness theorems with two propositions that states that not all mathematical questions can actually be solved this theorem was given by the well known austrain mathematician kurt godel in the nineteen thirties a new research by ben davids stipulates that machine learning is constrained by the exact same unresolvability a machines ability to learn which is known as learnability could be constrained by mathematics that is unprovable according to this argument simply put it is generally giving an ai an undividable problem which cannot be solved with a true or false response by an algorithm the team works on a machine learning problem in their research which they define as estimating the maximum in this problem a website aims to display targeted advertising to the visitors of the website who browse the site very frequently there is no certainity about which visitors will the visit the site though this kind of mathematical problem generally is similar to a machine learning framework called probably approximately correct learning it is also related to a mathematical puzzle known as the continuum hypothesis the continuum hypothesis is an other field of research for godel similar to the incompleteness theorems it is perturbed with mathematics that cannot be proved to be true or false machine learning could significantly run into a similar perpetual stalemate given the scenario of approximating the maximum example