I’ve always been fascinated by the idea of computers and speech. Many, many years ago, I was responsible for launching the Texas Instruments Professional Computer (TIPC) – TI’s response to the IBM PC – in the UK. It was a great piece of kit that was better than IBM’s in almost every way. The problem was, to achieve that ‘better’ it was also not IBM compatible which, given the dynamics of the market, meant that it sank without trace very rapidly.
One of the ways in which it was better was that it featured speech recognition technology – and this was back in 1983. I clearly recall hauling its sibling, the TI Portable Professional Computer (TIPPC) – and ‘hauling’ was the operative word, given that the color monitor version of the thing weighed 14kgs (31 lbs) – around to demonstrate it to customers who were, to a man/woman, amazed to see it respond to my voice. (I have to admit to some cheating: the demo was very carefully scripted so that the TIPPC appeared to recognize anything I said, which was very far from the case.)
And: which of us who saw 2001: A Space Odyssey can forget hearing the chilling tones of HAL saying: “I’m afraid I can’t do that, Dave”.
Fundamental limitations
I also spent a disproportionate amount of time seeking out voice recognition and voice-to-text software for my home PC. I recall Dragon being the best of a pretty lackluster bunch. The problem back then, of course, was the fundamental limitations of the hardware: as good a processor as the 8086 (on which the TIPC was based, with its standard 64k of memory, upgradable to a whopping 768k) was in its day, it was never going to be able to handle the software complexity necessary to recognize natural speech.
Fast forward 30+ years, and hardware performance is light years from where it once was – not only in terms of processor capability, but also the capacity, density, speed and affordability of memory. That’s given developers the almost unfettered ability to create the code they need to (or, for the more cynical of us, to indulge in the kind of software bloat that makes the latest version of MS Office so painful to use when compared with, for example, Office 2003).
One of the primary beneficiaries of this almost unlimited processing power has been AI. It too has been around a long time. A later version of the TIPC, for example, ran an expert system called “Personal Consultant” (and TI had a team of so-called ‘knowledge engineers’). Most famously, it was adopted by Campbell’s Soup to capture the knowledge of one Aldo Cimino, an employee with 46 years’ service.
AI can now leverage this almost limitless computing horsepower and is, of course, a fundamental underpinning technology for speech recognition such as that found not only in Amazon’s Echo, but also in Apple’s Siri, Microsoft’s Cortana and OK Google. Neural networks, deep learning, machine learning – all have a role to play.
Hugely attractive
But: how will all this play out in the military? Is speech recognition yet ready for mission critical applications that are a matter of life and death? It is, potentially, hugely attractive. Much development effort is being applied in recognition of the increasing operator workload in, for example, flying a jet fighter in a combat zone. Speech recognition can theoretically ease that burden, in tandem with other human/machine interaction technologies.
However: as advanced as, for example, Amazon’s Echo is, it’s a long way from flawless. It’s all too easy to accidentally wake Alexa up. Her comprehension, while still a thing of marvel, is still far from perfect. And: how well she would function in an environment characterized by significant ambient noise – such as that typical of a cockpit – can at best be described as questionable. There are also issues of security and integrity to consider.
Here’s also the problem of the intimate relationship between speech recognition and artificial intelligence, given the philosophical questions that are being asked about whether we really know what kind of genie we’re letting out of the bottle when it comes to AI. After all: which pilot wants to hear, as Dave did: “I know I've made some very poor decisions recently, but I can give you my complete assurance that my work will be back to normal. I've still got the greatest enthusiasm and confidence in the mission. And I want to help you.”
