There’s an ongoing argument in the tech community regarding whether advancements in AI are likely to be beneficial or harmful to humanity. Although they’ve previously staked out positions on the matter, in the past few days this has boiled over into a public spat between Mark Zuckerberg and Elon Musk.
While some commentators have said that this is simply a matter of the two protecting their personal brands, I don’t believe their argument is a conscious matter of marketing, and I don’t think it’s a fair evaluation of either of their points. Rather, I suspect that they’re simply looking at two sides of the same coin, through the filter of their personal experience.
From where Zuckerberg is sitting, AI is already used to make Facebook work better: to better match up content to users, to better allocate data centre resources. Every new technological advancement leads to him hiring more recent PhD graduates, better service, more efficient use of resources. He has also made a valid point with regards to self-driving cars saving lives (an aspect of the discussion where it is likely that he and Musk agree).
From where Musk is sitting, AI is likely going to take over vast additional areas of manufacturing, ultimately finishing off the process that automation and off-shoring started. He may personally gain in the short-term from the reduced costs of building product, but he knows he also has to sell to somebody – and if that person doesn’t have a job, they’re likely not going to be buying a luxury car (or a trip to Mars, for that matter).
The take-away will be no surprise to most readers: AI is disruptive. It will (and is already) benefiting some people, while causing obvious (hopefully, but not necessarily, short-term) harm others. It is impossible to determine right now whether there will be a net benefit on the far side of whatever societal disruption occurs. Opinion of public figures with regards to AI likely rests on whether they will personally benefit (whether they realize this consciously or not), and it is probably worthwhile to interpret their remarks that way.
There’s a lot of talk in the Bitcoin community about the possibility of a hard fork in the blockchain at some point.
The underlying issue has to do with the size limit of transaction blocks, which result in limits to the time it takes to process a Bitcoin transaction. There’s a good article on the philosophy of the two sides on QZ (link), which is worth reading if you need a primer.
Essentially, one side wants to double the size of a block from 1mb to 2mb, and the other side wants blocks to be unlimited in size. Both options have drawbacks (both technical and philosophical) that I won’t get into here.
I’d like to propose a compromise between the two options. Continue reading
I’m not sure if this is an original idea, but sharing just in case.
Many modern cars have push-button starters that do not require a key to start the ignition. Usually, the proximity of the owner’s fob is sufficient to start the car. The problem, of course, is that the code to open the car is the same, and (by design) it needs to work from a distance. This provides ample opportunity for hackers to intercept the code (and either open, or steal the car), even with various techniques that try to obscure the code.
It occurs to me that the process of opening and of starting the car do not need to be combined.
The fob could contain two transmitters, with different coding schemes. One of them would be used for remote entry, as in current designs. The other, which would be extremely low-powered, and only operate from a range of two or three feet, would be used to enable the ignition.
What has more strings than a Rock Ock, spins around like a whirling dervish, and sounds like a cross between a steel drum and an electric violin? Why, a spin-tar, of course.
Somebody stop me if this actually exists. There are a lot of obscure musical instruments in the world, and I’m not an expert. Continue reading
I wish I’d spotted this story (also on QZ) a few days earlier, as it would have been a nice one to add to my Five Year Retrospective.
The company implementing the Hyperloop demonstrator have announced $8.5 million in funding – get this – to build a freight-first version of the technology.
Called this one in 2013 (here) – “this system is wasted on human passengers”.
Seriously, are these people reading my blog, or is my real name “Captain Obvious”?
I realized today that I’ve been writing this blog for more than five years. I’ve kept everything up here (even the stuff that now appears hopelessly naive or even downright embarrassing), because it provides a record for me of the lessons (some of them hard earned) that I’ve learned along the way.
Re-reading some of what I’ve written, I’ve decided to write a retrospective, to see how well things have stood up over time (and to review what I’ve learned along the way).
The results are a mixed bag, as you’ll see below. Continue reading
I wrote about VW’s vaunted platform streamlining process about a year and a half ago (you can read my post here), and predicted at the time that it would eventually cause them grief.
I just read this article which describes how their platform is already backfiring on them. I had though it would take a few years for this to happen, but the auto industry evolves rapidly.
I’m not sure my reasoning was 100% accurate for why things aren’t working out there, but it still brings up an important point – a streamlining process can result in massive over-specialization based on the way things work today. That can have severe ramifications when the underlying model changes, because it can slow down the ability for an organization to change later on.
In addition, the actual process of streamlining can involve large organizational change, and as we all know, change management can be a tricky task.
Google just announced that it has bought Nest (producers of smart thermostats, and now smoke alarms too). Both sides (for now) are indicating that they’ll take privacy seriously.
Its clear that Google is chasing a strategy with the “internet of things”. Willing to bet they’ll make further purchases or create new products along these lines in the very near future – think connected burglar alarms (Rogers won’t be happy), smart door locks, or light bulbs that compete with GE’s connected product.
Its also clear that this is going to push other large tech competitors to do much of the same, for fear of falling behind. Expect much investment and many buyouts in this sector this year.
I recently read an interesting book called Dogfight, about the war between Apple and Google over the cellphone market. If you haven’t read it, it covers some of the background behind the development of the iPhone and Android, and the strategies that both companies employed to fight each other.
The major IT companies have long been converging, in the sense that their offerings overlap more and more over time, and the strategic maneuvering of the past few years has gradually made way for something more like trench warfare – gradually improving their products in the hopes of outselling the rest, while relying on the entrenchment of their own platforms. Obtaining a new customer in this environment is zero-sum – the customer must be leveraged away somehow from a competing platform. This is one of the reasons why I wrote some while back that I was bearish on big tech companies in the immediate future. So far, the results of my predictions have been mixed, but I think we’ve already seen them start to play out. Continue reading
I’ve been thinking about alternative power-trains lately. I believe I’ve come up with something that is at least partially novel, and that might have applications for nanotechnology (I don’t think it scales up, unfortunately).
The basic idea is a modification of a standard electric motor (linear or rotary – I’ll show examples with use-cases below). The thought I had was whether it was possible to make the “fuel” for the motor directly power it, rather than using electromagnets powered by electricity. Continue reading