This is a download from BFM 89.9, the Business Station. BFM 89.9, the Business Station. My name is Rich Bradbury. This is Matt Splaind.
What is it to be human? It's a topic we've come back to on and off over the years on Matt Splaind. Mainly because Matt is looking forward to the day when he can finally be a real boy. Recent advances in neurological treatments for stroke victims and breakthrough in artificial intelligence are once again converging to make Matt question everyone else's humanity.
You didn't want to do something simple like, I don't know, talk about the new Apple lineup or something. I mean, do you want to talk about the new Apple lineup? Yeah, exactly. I mean, it's slightly better phones, slightly better watch.
I like the multi-colored iMacs. It's a really cool knob to those turn of the century, turn of the Millennium G3 models. I can't actually use them. They've got those 4.5K screens which hurt my eyes.
The 5Ks are a bit better, but I'm better off with a Mac Mini. But it's very hard to find any devices these days from any of the major manufacturers that aren't great products. Gone are the days of companies trying new form factors or formats for their mobile devices. The closest we've got to something interesting is that new laughter, smartphones that fold, which is incredible from a technical point of view.
The fact that we have foldable touch screens. But the fact that so many people are wowed by phones that just replicate the form factor of the devices that we were using 15 years ago. Or the fact that Apple releases new computers that remind us of their old computers, that tells you a lot about where the industry is out of home. At this rate, somebody's going to suggest a smart device with a hardware keyboard attached.
And we'll be falling over ourselves because somebody's just reinvented the BlackBerry. Well, that's a good place to start. The word plenty of people who thought that the BlackBerry was the end of civilization. People with their heads and their devices and your trans human shows that tend to go in a similar direction.
Well, one of the reasons I've been thinking about this is because a friend was asking about the Metaverse, which is something that I think we talked about three shows ago, MSP176, which is really looking at a similar area, but from a different angle. So when you look at this from the virtual reality perspective, it's the idea of becoming a digital person, which is in a sense a different form of transhuman. You're letting go of your physical form in a way and interacting with people in a virtual environment. Yeah, but you're not actually letting go of your physical form, your physical movements are being rendered digitally.
Well, I get that. But even that's more of an interface thing, you know, even when we get to the point where those VR interfaces can translate brainwaves, then there'll be no need to wear that VR suit, there'll be no need to use a joystick, there'll be no need to run in the real world. You know, we can comfortably slump into that world that Bruce Willis showed us in the movie, Sariqet, you know, lying pretty much comatose while our minds inhabit machines. Now, of course, in that movie, the machines were actually physical.
They were robots that interacted in the real world, but in the Metaverse, it's much more likely that we will all be, you know, these digital sprites. Not to be argumentative, but it still isn't transhumanism, the strictive sense is it? Well, perhaps not, you know, I know that not many people have watched the sci-fi teen series, the 100, which isn't really a teen show at all, or at least that's what I tell myself. One of its sub-blocks involved a supercomputer that wanted to save mankind by uploading each person's consciousness to a digital simulation.
And it did that by attempting to destroy the planet and informing everyone that their physical bodies were about to be incinerated. You know, obviously, that's a bit harsh on all the animal species that inhabit the planet. But, you know, that still is transhumanism in a sense, because it's humanity transcending, you know, these meat suits to live in a perfect eternal life. It sounds like something that you, in fact, it sounds exactly like something that you kind of advocate.
Well, I actually quite like my meat suit, even though, you know, it often doesn't sound that way. Although having it enhanced would be quite useful, you know, I'd quite like a telescopic arm. Pneumatic legs would be pretty fun too. But when we've talked about human augmentation in the past, it's mostly been theoretical.
For example, computer brain interfaces that are designed to help people with dementia or devices that help people with mobility issues by essentially enhancing, repairing or replacing the faulty wiring in the bodies. So, things that are experimental or in development, rather than products that are market ready. But that's really what I want to start with today. With a breakthrough, that's helping stroke victims to regain their nobility.
Yeah, so I sourced the background for this from a Wired story called A Stroke Reveals The Future of Human Augmentation. We'll put the link to that in the show notes, which in turn took its cue from a study published in the Lancet earlier this year. It concerns a trial undertaken by the USC Neuro Restoration Center in California in 2017. Some reason, I find it very hard to say Neuro Restoration.
108 stroke patients who had lost mobility in an arm or a hand were implanted with a neurostimulator in the vagus nerve. Now, the vagus nerve is a cranial nerve that sends information from the brain to your heart, your lungs, your abdomen. And the implants stimulated the nerve and the subjects were also given physical rehabilitation as well. The results were pretty striking.
They all made significant progress in restoring mobility to the affected hand or arm. Even the people who hadn't responded to any previous treatments or therapies or interventions in the past. So, in effect, this is the patient's themselves learning how to reuse that affected limb. Yeah, so it's not a computer chip or an artificial limb control by the brain.
It's an electrical charge that's stimulating a nerve and that causes the brain to release more of its own neuromodulators. And that regulates how our body responds. So, you have a flood of these new components, if you like, at the time when you're doing that physical therapy. So, you start to see a much more rapid improvement in motor skills.
So, essentially, you're probably just doing what the physical therapy alone might have achieved eventually, but you're doing it on a much faster timescale. I think, wide-phrased, it's something like compressing years of physical therapy into a few months. Or if you want to put it into the language of a healthy person, training. So, this is something that a healthy person could use to hone or improve physical skills?
Well, yes. As the leader of the study, Charles Liu pointed out, it's not really any different to practicing a golf swing. So, you're trying to improve your motor control, your coordination. So, the neuro stimulation floods your system with neuromodulators and it makes your brain more amenable to learning or adapting to that new task.
Many of us have had that moment of practicing something over and over, maybe for weeks and months, and suddenly you make what you feel is a breakthrough. And it's what a lot of people call muscle memory. Obviously, it's not muscle memory because your muscles don't have a memory. It's your brain hard-coding that set of actions and recalling them.
And as the study designers admit, there's no real difference between teaching a stroke victim to lift a fork to his or her mouth and improving a tennis player's backhand. It's essentially the same motor reinforcement process. I mean, this is interesting. So, this could be the next debate in stuff like professional sports, whether this kind of augmentation is ethical.
Yeah. I mean, let's be clear. I mean, all the neuro stimulation in the world isn't going to turn me into Emma Radinkano. Sorry.
Thankfully. Well, mainly because I'm not an 18-year-old woman, and because I start from a base point of being old, unfit and awful at tennis. So, it's not going to give you a special aptitude, but it will help you to be the best tennis or badminton or squash player that you possibly can be. And it will allow you to get to that level faster.
And of course, that in itself is an enormous edge in professional sports. You can achieve your peak much earlier and probably as well with fewer injuries. So, that alone can allow you to have a longer career, and therefore you can achieve or earn more over that longer duration of your career. So, if you're not adding anything that isn't natural, is there even an ethical issue?
Well, that's one of the reasons that sport is a good place to frame this debate, because it is something that's highly regulated. Charles, for you, the study lead, he maintains that sporting bodies should be considering this now. At the moment, these are still semi-experimental treatments, because they require a surgical implant. But this is technology that could quickly become scalable and a lot more mainstream.
So, we already know that people with people learning many skills, whether they're athletes, musicians, they're planting more, those that are wealthy or have state or private resources backing them already have a considerable advantage, because they have access to professional trainers, they have better equipment, they have remedial treatments, where they're needed. And this might become another advantage available to those with deeper pockets? Well, yeah, especially given the areas that this kind of research is heading in, I mentioned that scalability. There's also a lot of research being done into non-invasive and magnetic methods of neuro stimulation.
So, an athletics body might be able to search for evidence of, say, a surgical implant. But what if there was no surgery? How would you test to see if a nerve had been electrically stimulated, or if an implant had been removed months or years before? In any case, even if there was an implant, does this count as cheating, or is it simply training?
I guess the larger conversation is about it's role in wider societies. Sports bodies will make their own mind up. But what about enhancing people like you and me, ordinary people? Well, this is why the debate is so interesting.
It's that word ordinary people. This is the trope that runs through superhero fiction, Peter Parker, rather than Clark Kent, where an accident makes an ordinary person extraordinary. So, do enhancements like this, or will enhancements like this, make some people extraordinary? Is this kind of stimulation limited to improving just motor skills?
Well, you see, you can imagine a world where all of the sports people use this as part of their training or all the professional musicians. Use team is also researching its efficacy on people with Parkinson's. And the treatment is showing some effectiveness in improving language skills. So, what if a small and relatively wealthy nation, say a Singapore or a Switzerland, if you want something that's maybe less regionally challenging, made this kind of treatment available to all of its residents?
And the kind of economic advantage of this might confer on them as a nation? Yeah, especially compared to some other nation that lacked the resources to follow along, or if we look at it at that more individual level, do we end up with those two tier societies, people who are ordinary and people who are extraordinary? You know, you go back to that superhero fiction. One of the storylines in X-Men is about society turning on the mutants and trying to eradicate anyone who has abilities over and above what we consider to be normal.
But at the other end of the spectrum, you have the watchmen, you have Dr. Manhattan who becomes so extraordinary that he's prepared to wipe out the species he evolved from, humanity. Of course, we're not talking about taking anything to anywhere near that kind of extreme, at least with the kind of Eurostimulation technology we have available. But it marks the realistic start of that first generation of externally enhanced humans.
Fascinating stuff. And when we come back, the machines of that code themselves, you're listening to Matt's Blaine here on BFM 89.9. Before Friday materialises, BFM 89.9. BFM 89.9, the Business Station.
My name is Rich Bradbury. This is Matt's Blaine. We're looking at real-life superheroes on Matt's Blaine today. People using technology and medical advances to give them professional and economic advantages over their normie friends.
One area I know you like to look at when we discuss transhumanism is the flip side of the coin. Machines that are becoming more human. Yeah, I mean, that's one of my kind of favourite subjects. We talked in the past about people who are experimenting with biohacking, for example, essentially doing DIY experiments in manipulating their own DNA, which is as dangerous as it sounds.
But often when we've talked about transhumanism in the past, it's been in terms of machine additions, not necessarily that robot arm that I said I quite like, but more along the lines of a chip in the brain to enhance memory or some other cognitive function. We even talked about the possibility of an advanced and implanted AI version of Siri or Alexa becoming kind of a part of your consciousness. So blurring those lines between the man machine as Kraftwerk put it, and one of the limits we put there in the advanced towards more capable and possibly sentient machines is that ability for machines to design the subsequent and next generations of machines that follow them. In other words, an AI that writes a code for its replacement.
Yeah, so you know, one area we've dived into before is the divergence of machines and humans, that as the machines become more complex, it tests our ability to understand how they work. And let's face it, most of us have no idea how the chips on our smart devices work as it is. So we reach a point where the only way to advance our computers to advance our AI is to get the machines we no longer understand to create a better and more complex machine and so on and so on. And from that point onwards, we have to trust that either we have some kind of kill switch that turns everything off or that we can trust the machines to do their best to work in our interests.
Which requires it to have some kind of critical thinking ability. Yeah, and that's something that we're definitely a very long way away from. You know, I talk about the impending jobs crisis that machines may bring about. And I get pushed back from people who rightly observe that AI has so far tended to enhance people's roles rather than replace them, that the machines do a lot of the drudge based crunching and grunt work.
But that often overlooks the fact that those machines are still replacing workers. They're not getting rid of them entirely, but you now need fewer human workers to do what that department did before, because that grunt work used to be 80% of our time. So companies need fewer humans to do that work. And usually that grunt work was done by people in entry level positions.
So you're removing one of the mechanisms for recruitment and training. And I think we've mentioned this before as well, using the ad industry as a model. It could mean creative agencies where AI handles much of the basic design work. And a creative director selects and guides the output that gets shown to the clients and becomes the finished work.
But how do you train the next generation of creative directors if there's no ladder left for people to get onto and decline? I'm not entirely sure that the world would mourn the extinction of creative directors. Perhaps not. I mean, it's just an example.
But we see similar patterns in industries like law and accounting, where there's a lot of form filling, there's a lot of repetition. But that form filling and repetition helps to hone the abilities of lawyers and accountants. The machines don't always get things right because of that issue about context and nuance. So you still need that human or humanish expert to step in and make sure that your interests are being best served.
But without those gateway positions, we may not be able to adequately train that next generation of people that we will rely on to become experts. So your answer is more and better AI? Well, it's one of those questions that we really can't answer at this point. I mean, I have made that argument in the past.
But that's why we're at this interesting tipping point with AI in terms of determining what it should and shouldn't do. And how we decide to limit its capabilities. And like the human augmentation that we spoke about before the break, the time to be looking at those parameters is now because we are approaching that point where we will be guided by the realities on the ground and taking a reactive rather than a considered response based on the fact that the technology is already here. Oh, we have the point where machines can create code, machines that can create their successors?
Create their own successors, thankfully, no. But we do have AI that can write its own code. So codecs, which is built by OpenAI, is pretty much a happy accident from the natural language processing AI GPT-3, which is something we've talked about in the show before. So GPT-3 can write articles, social media posts, even poetry from scratch.
You tell the machine what you want an article or a poster to be about, and it will generate it, as I said, from scratch after trawling the internet for background information. So it's not selecting and passing phrases, which you might expect it to be doing, you know, just doing a copy and paste from what it finds online. It's generating entirely new text and passages based around the subject. And because its neural network was trained on all sorts of text posted on the internet, as well as digital books and other information sources, it unexpectedly developed the ability to code simple programs.
And codecs is an evolution of that ability with GPT-3? Yes, it's a dedicated coding AI. But because of that underpinning, it understands code, and it also understands natural language commands. So you can tell the AI to write your program by giving it some code prompts, or simply tell it to write your program that can check for valid postcodes in a database, for example.
Now, I know that's not the most exciting scenario for a program, but it's a natural example that the New York Times set up for codecs to execute. And it created the program, and it started spitting out the responses in a few seconds. So then we go back to that natural language underpinning. codecs can write programs in 12 different coding languages, but it can also translate between them.
And that's immensely useful, because it means you can then adapt the same program for different environments and systems automatically. And what are the limitations or the restraints of this system? Well, it's not brilliant. OpenAI estimates that the system, as it currently stands, will produce the right code around 37% of the time.
Some of the time the code simply doesn't run, it might have security vulnerabilities in it. And that's a topic that we'll come back to at the end of the show of this time. But very often the code needs to be checked by someone who really knows the system it's designed to run on, a coder. Maybe there are small tweaks that need to be made because codecs hasn't fully understood the architecture it's designed to operate on.
Or maybe it was something as simple as information that was missing from the inputs that the, sorry, or maybe it was a sim, or maybe it was something as simple as information that was missing from the inputs that it was given at the start. So essentially, this is more machinery doing that grunt work we're talking about. Definitely. I mean, this is a developer's aid rather than something that replaces the developer, but it's hoped that it will even help the next generation of coders to learn.
So GitHub, which is home to many open source coding projects and repositories, is integrating codecs into a tool it calls co-pilot, which suggests the next line of code in the same way that text tools on our phones and computers help to auto-complete our sentences. So on the one hand, you know, it speeds up the process for experienced coders. But on the other, it's a way for newbies to create something that works, or nearly works as is the case in that non-37% of the time. So we're not in any danger of sentience breaking through at this point of a machine creating something so powerful and unknowable that we have to hand control of it over to other machines.
No, not yet. But you can see the direction that we're heading in. It's going to be interesting to come back to codecs in 12 months and see whether that 37% success rate has become 50% or 90% or 90%, or whether it becomes stuck at a certain point of effectiveness or accuracy. Then we'll start to see whether this is because of the limitations of its own programming.
But that doesn't mean that we shouldn't be talking about how much we want machines like this to be able to do in the future, and what limitations we should start to deliberately build into their programming. You know, I'd like to see sentient machines in my lifetime, but it does look increasingly unlikely. Okay, we've got time and you said that you were going to come back to some of these security vulnerabilities. Talk to me about that.
Yeah, so this is a little bit off topic, but we've recently heard about zero-day vulnerabilities that hackers have been able to use to hack into our mobile devices and computers. A lot of it is centered around spyware that the Israeli company, NSO Group, has sold to governments around the world. So even if you don't think you're at any risk of surveillance by any particular government, zero-day flaws are ones that operating system manufacturers are not aware of. That means that commercial Black hackers may also have spotted the same ways to target and get into devices.
And we're often a bit slow and lazy when it comes to installing those OS updates, especially when, you know, the security fixes rather than something that gives us a new way to use that device. But now is actually a really good time to click yes on that notification and update. So this week Apple, Microsoft and Google have all published security updates that address some recently uncovered zero-day weaknesses in their software. So if you're on Windows or if you're using iOS, don't delay update today.
And the same, if you're using a Chrome browser, it's time to update as well. Google has released an update for the browser, though it's with holding information on how the vulnerabilities it's found might be used until a majority of its users have updated, presumably because they want to avoid tipping the hackers off when there are still plenty of users using older versions of the browser who could potentially be targeted and exploited. So then I've actually ended a show by saying something useful. Thanks very much for that Matt, great show.
Thank you, pleasure to be here as always. Now you can find Matt on Instagram and Twitter over at culture. You can also head over to culturepop.com for transcripts of these shows and information on culturepop and its consulting services. And if you did miss any part of the show, you can download the podcast wherever you normally get it from.
I suggest using the BFM app. It's available at Apple App Store or Google Play. This is Rich Bradbury for BFM 89.9.