China’s Belt and Road initiative has not stood the test of time + One thing AI can’t do yet: Logical and social reasoning

Cracks in several of China’s Belt and Road infrastructure mega-projects have come to light: Over the last 10 years, China has disbursed USD 1 tn in foreign loans, as part of the Belt and Road initiative — intended to develop economic trade and expand China’s influence across Asia, Africa and South America— making Beijing the largest government lender to the developing world, the Wall Street Journal reports, citing World Bank data. Now, some of these projects have come under fire for their poor construction, which risks obstructing important infrastructure and burdening countries with additional expenses — and heavy debt burdens — for several years as they attempt to address their issues.
One major example: Ecuador’s main source of power is plagued with construction issues: The USD 2.7 bn Coca Codo Sinclair hydroelectric plant in Ecuador — the country’s biggest infrastructure project — is at risk of breaking down and its dam could potentially be damaged by the erosion of surrounding slopes. “We are suffering today because of the bad quality of equipment and parts” in Chinese-built projects, said René Ortiz, Ecuador’s former energy minister.
Cracks have also emerged elsewhere: The Neelum-Jhelum hydroelectric facility in Pakistan was shut down last year after inspectors found cracks in the tunnel that carries water through a mountain to power a turbine. Uganda’s power generation company said it has identified over 500 construction defects in a Chinese-built 183 MW hydropower plant on the Nile that has experienced repeated breakdowns since it began operating in 2019.
Different training methods are required to advance AI softwares in logical and social reasoning, researchers suggest. Although AI appears to write well-constructed sentences, it is incapable of processing thoughts to generate accurate arithmetic and spelling, nor delivering accurate ways to do simple tasks like frying an egg, The Atlantic reports. MIT cognitive neuroscientist Anna Ivanova has shown, through brain scans, a clear separation between language and thought neurons — which is why patients with aphasia, suffering from the loss of the ability to understand or speak from brain damage, can still utilize their thought neurons and excel in tasks such as math, music, and coding.
Given that AI is permeating every aspect of our lives, softwares such as ChatGPT, with language mastery, must upscale their models to “navigate logical and social context,” to be able to avoid mistakes, including disseminating false information, says Kyle Mahowald, a linguist professor at the University of Texas at Austin, and co-author of various linguistic papers.