The impact on language in the era of intelligent chatbots and neural implants.
KEY POINTS:
• With the advent of neural implant technology, the prospect of tech-enhanced
humans, within our lifetime is a real possibility
• This raises the prospect of language being streamed, rendering language
learning obsolete
• But if language becomes a commodity, this raises dangers as well as
opportunities
Dire predictions about artificial intelligence surpassing and even supplanting humans has long been a staple of science fiction. These fears were originally highlighted in Isaac Asimov’s famous Robot Series of stories and books, originally in 1940, which gave the world the Three Laws of Robotics—the first attempt to address the ethics of artificial intelligence.
In the twenty-first century, with the emergence of a Fourth Industrial Revolution—sometimes called 4IR—science is catching up with fiction. Today, rapid developments in technology are leading to smart automation, and ever increasing interconnectedness between devices, and even humans.
Fears of the potential impact of 4IR on society led, in 2015, to many of the world’s leading scientists (Stephen Hawking among them) to publish an Open Letter, and a report, addressing these developments. The Open Letter called for concrete steps to ensure that as AI develops apace, controls are required to avoid negative societal impact, ranging from loss of employment caused by automation, to longer-term existential threats.
But in November 2022, with the launch of OpenAI’s ChatGPT, artificial models of language now arguably surpass human linguistic and intellectual abilities. And as language is, I have argued at length in several of my books, the hallmark of what it means to be human, in the third decade of the current century, we appear to have reached an inflection point, in terms of AI versus human.
One solution to the potential threat from AI is to use technology to hybridize the human mind. With medical advances, such as the Elon Musk-led venture Neuralink, implantable neural chips developed to create ‘smart’ brains will, in due course, receive licences for human testing. This raises the prospect of humans being able to communicate directly with the Internet of Things. Such a development will inevitably, in a matter of decades, change how we live our lives, perhaps even rendering the need to learn language redundant.
Computer-Brain-Interfaces and neuroprosthetic technology Non-invasive computer-brain-interface research and technologies were pioneered in the early 1970s. The goal of this technology attempted to “interpret” the electrical signals (“neural code”) produced by the brain. The principle was to enabling subjects with physical (motor) or speech disabilities to communicate directly with external interfaces, thus improving quality of life, diagnostic capabilities, and so on.
But today, a new era of neural implants, aka neuroprosthetic technology, has the advantage that it might, theoretically at least, both allow communication and control of external devices—by the power of thought alone—as well as receiving external signals to bypass or repair brain damage, and even enhancing native human abilities. This goes well beyond the assistive technology famously used by Stephen Hawking to communicate.
Hybridization of the mind: Creating the transhuman From a medical perspective, neural implants wired to various areas of the brain solves offers medical solutions, event to recalcitrant challenges, such as restoring vision to the congenitally blind, by correcting or bridging defective neural pathways. Such a claim was made most recently at Neuralink’s Fall 2022 Show and Tell presentation.
But arguably even more eye-catching, are claims that within our lifetime, technology will exist to enhance everyday life, by hybridizing the mind—giving rise to transhumans: tech-enhanced humans with new abilities.
One such case is language. Elon Musk himself predicted, in 2020, that a neural implant will, within the next 5-10 make language learning obsolete. The idea would work something like this: a language chip, implanted in the brain, paired with a wifi transceiver, perhaps inserted behind the ear, would allow us to stream language on demand, any language. At a stroke, speaking a new language would only be constrained by how many bars you have on your wifi signal.
The 'hard' problem of language streaming technology Science does now show that electrical activity for language in the brain can be decoded and interpreted (so-called voicegrams), and today these patterns of electrical activity can even be used to produce artificial or synthetic speech. This offers the prospect, at the very least, that subjects who cannot speak, can use the power of thought alone to produce language, using an external speech synthesizer.
Nevertheless, the prediction that language learning will become obsolete in a decade is wildly overoptimistic. There are many, significant challenges involved.
First, neuroprosthetic technology will need to be able to effectively communicate with the two brain areas that process language. It will also need to be able to communicate all the areas of the brain where concepts derive—the ideas that we use language to encode and externalize. As these concepts are found literally everywhere—visual concepts arise in visual cortex, emotion concepts in the amygdala, and so on—then the language chip would need to be connected with most areas of the brain.
In addition, the language chip would need to be able to receive on-demand streaming of the symbolic code that makes up any given language—words and grammatical patterns—which the language chip could decode from wifi data packets, transmitted from an external source, such as a language database. And these would be called up via the streaming signal, based on the relevant concepts arising in different brain areas, so that an individual can stream and so produce language and communicate effectively in real time, as they think, without having to actually learn a native language.
The Babel Apocalypse For now, this sort of language streaming technology remains in the realm of science fiction rather than science. Nevertheless, this is a future that may be closer than we think.
In my forthcoming book, The Babel Apocalypse, a work of dystopian fiction, I examine what such a future would look like. In particular, the book makes the two following predictions:
The ‘hard’ problem of making language learning possible, via language streaming technology, will exist by the end of the century. And second, voice-command ‘voiceprint’ technology, encoded in spoken metadata, will enable unique identification of individuals as they speak, in real time, which will transform security protocols: voice-commands will allow an individual to securely gain access to homes, offices, vehicles, retail and bank accounts, rendering keys, passwords, obsolete.
But such language streaming technology brings with it clear dangers. Here are three significant ones:
· There will be significant societal, ethical and civil liberty implications of language streaming technology, as individual versus state interests (in terms of population registration, crime agencies) will need to be carefully balanced.
· Safeguards in terms of legislation will need to be carefully calibrated to ensure potential overreach from big tech does not jeopardize individual freedoms when language becomes a commodity, ‘owned’ by tech companies.
· There will be cultural and identity implications: how is language regulated, when language change is a function of corporate management (with shareholders) rather than an organic, linguistic-community endeavour? How do we avoid something similar to the big tech-related ‘censorship’, currently the case for Emoji, which is controlled by Unicode based in California?
But one thing is becoming clear. The future of the mind is hybridization through technology. And if, one day, we no longer learn language, but stream it, just like we do music and movies, what does it then mean to be human?