In a year that has seen all of us adapting to new models for conducting virtually every aspect of our lives (virtually), one thing that has seemingly never failed to carry on the same as ever is artificial intelligence. 202o has been a major year for language AI, in particular, and the stories from throughout the academic and business worlds continue to astonish us with what is proving possible in communications without human input.
Week in and week out, stories of all that humans are enduring in 2020 appear beside stories of how AI is topping new hurdles and helping in these challenges as never before, from the bleeding edge of academic research to real applications that have shaken up medical research. Is it any surprise that an intelligence beyond the reach of an infectious virus should prevail unfazed while people struggle, or does machine learning simply happen to be entering a breakthrough phase at the very moment when there is virtually everything else imaginable to distract from it? For a better idea, we decided to look back at the year in AI as CSOFT witnessed it.
There are advantages to avoiding the scrutiny that comes with disquietingly powerful leaps of power and capability if you happen to be working on these kinds of things in an otherwise tumultuous year. Even so, language AI itself seems to be generating less fear mongering than it was just a year ago despite the startling progress it is making. Around the web, jaws seemed to drop at the news that OpenAI’s GPT-2 natural language processing algorithm has a successor with even more startling capabilities, but others were quick to highlight that even when GPT-3 making up dark fiction, it doesn’t really what it is doing. (We found it particularly interesting that the AI enjoyed some time on Reddit as a bot, undetected by any of the human Redditors in its midst.)
Departing from the familiar talk of what follows the singularity, 2020 saw an infusion of discussion around ethics in AI from a perspective of preventing biases – often human – from prevailing in its determinations about people. Much has been said about the arc of these concerns at major tech enterprises, beginning from early signs of a push for inclusivity on the part of the developers of these technologies themselves during the first half of the year. As companies struggle with the discovery that their more mature AIs have grown up on biased input from humans, this will likely be a major focus area in the year to come.
Most of the essence in 2020, however, is how AI has actually become a force for change and progress in medicine, healthcare, and the global pandemic response. At the start of the pandemic, many noted that an algorithm had very accurately predicted the likelihood of such an event. From there, we looked with fascination at how AI became a powerful tool in the global pandemic response, particularly with respect to expediting research and sorting global data for access and use in these efforts. As if this was not astonishing enough, news came just in time to cap the year off that DeepMind’s AlphaFold had demonstrated a readiness to model how a protein should fold at the atomic level, doing in hours what people might not manage in years. All of this is to say that where AI excels – spotting patterns and details that are too grand or granular for people to grapple with – has earned it a great deal of validation in the past twelve months. Looking forward to 2021 and what appears to be the final phase of the push to contain COVID-19, we are optimistic that AI will continue to surprise us with new ways of being helpful.
CSOFT works closely with technology and language AI solutions to deliver their products and services across borders and languages through rapid, accurate, and culturally-appropriate multilingual localization solutions. With our worldwide network of linguists and subject matter experts, CSOFT can help leaders in artificial intelligence translate their products and services for a broader range of users and enter new markets at the pace of technology today. Learn more at csoftintl.com.