Over the past few years, the unrestrained use of poorly-understood AI in real-world applications had inflicted significant harm upon #social, #economical, #educational, and #democratic infrastructures, round the globe.
Throughout AI's evolution—from the early 1940s McCulloch-Pitts neuron model, through the mid 1980s NN, to the early 2020s DNN—this technology had always been accessible to academic researchers and independent practitioners, in the form of peer-reviewed scientific papers, curated data sets, open competitions, and even open-source implementations. The intellectual insights of the academics and the social concerns of the independents had always counterbalanced the excesses of the industry. But all that had changed, with the advent of the post-Transformer Foundation Models (FMs).
FMs are conceived, built, deployed, and evolved exclusively by BigTech, behind the veil of commercial secrecy and under the tight control by a few billionaires. Academic institutions and independent laboratories no longer have access to the data, software, hardware, procedures, and policies that govern these FMs, which are now re-quilting the World's socio-economic fabric, in real-time.
#AI now poses a clear and present danger to fundamental human rights. It must be #regulated, uniformly, globally, with deliberate haste.