Joseph Wilson, a linguist and journalist who has done considerable work with oral languages (languages not yet written down), offers some thoughts on claims that chatbots like Blake Lemoine's LaMDA, really speak like human persons. But this excludes all unwritten forms of communication: sign language, oral histories, body language, tone of voice, and the broader cultural context in which people find themselves speaking. In other words, it leaves out much of the interesting stuff that makes nuanced communication between people possible. We really don't know how old spoken language is (Wilson suggests 50,000 years) but written language can be traced only as far back as about 5400 years ago. And only about half of all languages (he estimates 7100 currently) have ever been written down.

AI-powered speech translation has mainly focused on written languages, yet nearly 3,500 living languages are primarily spoken and don't have a widely used writing system. This makes it impossible to build machine translation tools using standard techniques, which require large amounts of written text in order to train an AI model. To address this challenge, we've built the first AI-powered speech-to-speech translation system for Hokkien, a primarily oral language that's widely spoken within the Chinese diaspora but lacks a standard written form. We're open-sourcing our Hokkien translation models, evaluation datasets and research papers so that others can reproduce and build on our work. The translation system is part of our Universal Speech Translator project, which is developing new AI methods that we hope will eventually allow real-time speech-to-speech translation across many languages.

Text-based language models such as BERT, RoBERTa, and GPT-3 have made huge strides in recent years. When given written words as input, they can generate extremely realistic text on virtually any topic. In addition, they also provide useful pretrained models that can be fine-tuned for a variety of difficult natural language processing (NLP) applications, including sentiment analysis, translation, information retrieval, inferences, and summarization, using only a few labels or examples (e.g., BART and XLM-R). There is an important limitation, however: These applications are mainly restricted to languages with very large text data sets suitable for training AI models. GSLM leverages recent breakthroughs in representation learning, allowing it to work directly from only raw audio signals, without any labels or text.

We are an amalgam of genes, knowledge, culture and religion, which have influenced our lineage for thousands, tens of thousands of years. One thing is linked to the other, and the history of humanity could not be explained without physical evolution, religion, politics, or science. We are different from our ancestors who lived in the 19th century, or in Prehistory, because we have evolved in different ways. But, What is causing the great changes in humanity? A conquering king, a pandemic, a new religion can change the destiny of hundreds of millions of people. But, objectively, the most important evolutionary changes have come from the hand of technology.

"[[[When in the course of human events it becomes necessary for one people [to dissolve the political bands [which have connected them with another]] and [to assume among the powers of the earth, the separate and equal station [to which the laws of Nature and of Nature's God entitle them]]], a decent respect to the opinions of mankind requires [that they should declare the causes [which impel them to the separation]]]." But how did it ever make its way into the world? At 71 words, it is composed of eight separate clauses, each anchored by its own verb, nested within one another in various arrangements. The main clause (a decent respect to the opinions of mankind requires …) hangs suspended above a 50-word subordinate clause that must first be unfurled. To some linguists, Noam Chomsky among them, sentences like these illustrate an essential property of human language. These scientists have argued that recursion, a technique that allows chunks of language such as sentences to be embedded inside each other (with no hard limit on the number of nestings) is a universal human ability, perhaps even the one uniquely human ability that supports language. It's what allows us to create--literally--an infinite variety of novel sentences out of a limited inventory of words.

Neuralink – which is'developing ultra high bandwidth brain-machine interfaces to connect humans and computers' – is probably a bad idea. If you understand the science behind it, and that's what you wanted to hear, you can stop reading. But this is an absurdly simple narrative to spin about Neuralink and an unhelpful attitude to have when it comes to understanding the role of technology in the world around us, and what we might do about it. Neuralink is developing a'whole brain interface', essentially a network of tiny electrodes linked to your brain that the company envisions will allow us to communicate wirelessly with the world. It's easy to be cynical about everything Silicon Valley does, but sometimes it comes up with something so compelling, fascinating and confounding it cannot be dismissed; or embraced uncritically. Putting aside the hyperbole and hand-wringing that usually follows announcements like this, Neuralink is a massive idea.