Why AI needs to be considered carefully.
Let’s play ‘AI Bingo’ – talking about Artificial Intelligence? Picture of Terminator….check!
There’s lots of talk (and quite rightly so) at present surrounding the advancements of Artificial Intelligence. (a great article here) How will this impact our day to day lives? Will it mark the end for human life as we know it? Will we live in a Terminator style world where we are merely slaves to the machine enclave as it fires nukes at us to eliminate us as a threat?
Let me be absolutely clear to those who are not sure. AI (and its advancements) is not about a 6ft robot walking around acting like a human and talking like a human (with a slight German twang). That’s an obvious technological enhancement that we probably won’t see soon, if ever.
What AI is, (and let me be clear I don’t think many people realise how advanced AI is), AI is banks of ever powerful and ever scalable servers sitting in bunkers in mountains with eyes and ears everywhere. AI is these servers being stood up by somewhat unknown tech companies gobbling up data every second. Plugged into cameras, CCTV networks, Military, Smartphone microphones, webcams….AI is being able to augment all this information with other sources of big data & make tangible assumptions and decisions based on it (the data), quicker than any human could.
AI is about machines extending their utility function.
AI is a massively global social network platform, gobbling up our inner-most thoughts and data, pictures and blogs, our family ties & our instant messaging all containing sensitive info.
Think about all these Amazon Echo and Google Home Assistant devices we’re installing, they are all listening & cataloguing everything we say. Our home automation devices track our movement, our CCTV cameras dump our movements to the cloud – even our health data from FitBit can be accessed and augmented with other data to build up such an accurate picture of each of us.
Do you honestly think your smartphone microphone isn’t on right now? Really? What you happen to be talking to your colleague about Narrowboats & all of a sudden your Gmail client is serving you sidebar ads for narrowboating holidays or waterway information! Coincidence? Or what about your voice activated smart TV – only listening when you ask it to switch channel? Your Xbox One politely covering it’s ears up when you’re talking to your family & loved ones about something sensitive. Yeah, I don’t think so.
GHCQ were tracking physical copper telephone wire, the NSA were red-flagging human conversation from encrypted cellular phones and satellites were watching our every move 60 years ago! Imagine what we’re doing now.
This type of intelligence is growing exponentially and that forms the true heart of what AI is and any inherent risk. Let’s take an example, if we were to ask an AI how to eliminate SPAM? Surely the most obvious answer would be to eliminate the source/creator – so that’d be us then, us fleshly humans? Right, let’s fire that nuke. That’s a pretty simple example of computers expanding their utility function out of context and applying an answer which would risk the human race. But it’s feasible. It’s very feasible.
Some of the points I’m making here are mirrored by a talk I watched from Elon Musk about the very same subject. And he’s right – although I do think it’s somewhat ironic that someone whose wanting to warn the world about being mindful of the dangers of AI or in his words ‘summoning the demon’ is also responsible for building one of the biggest self-learning networks in existence that continues to grow today.
If you don’t know, TESLA motor company runs each car as a ‘node’, in effect a PC on a network equipped with a myriad of sensors from cameras to environment sensors, tracking, GPS, you get the idea. To you it’s just a car. To a network these are just sensors in the matrix. To a tech company (and let’s be clear TESLA is a tech company first & foremost) this is just big data.
Each car sends all this data (including your user journey information) straight back to TESLAs central network (think of it like the Borg in Star Trek) and it then collects this information from all its cars & passes it around back to the cars to help them “learn” exponentially. The more cars sold, the more sensors brought onto the network, the faster each learn and the more powerful the network becomes through this collective ‘hive’ mentality. We are many but we are one etc.
This is really fucking dangerous in the wrong hands.
Just think about that for a second. Now think about the dangers of a computer making a decision to for example, run cars off the road at high speed where sensors had identified that the occupants were exhaling a poison (alcohol or cannabis for example – TESLA cars can come equipped with oxygen generators and toxicology sensors) and decided this would be a suitable punishment. Sounds far-fetched?
What if one TESLA car with occupants who’d consumed alcohol were on a head on collision with another TESLA that had children in it? Who makes the decision? A machine? I don’t want to have this blog go into self-driving car issues in particular but you get my point about the dangers of AI right?
And I’m an advocate! Not only of AI advancement but also of people like Elon Musk who seemingly are at the forefront not only of that type of advancement but seem to want to have these important conversations.
But we do need to tread carefully & ensure people who are adequately qualified are calling the shots.
So it gets my goat when people like Musk have to leave councils to help luddites like the POTUS because people like that ‘don’t get it’ but are some of the most powerful people in the world. I preferred Obama in this respect, at least he was asking intelligent questions.
Regulating the Internet
In a somewhat similar technology topic, I read this week that Theresa May’s approach to counter-terrorism and getting ‘tough’ is to demand a ‘regulated internet’. This is a statement coming from someone who probably fucking struggles to program her Sky Q.
Regulating the Internet means what exactly Mrs May? I’m not sure it’s good for our planet that people like you get a say in things you don’t know anything about. Its decisions people like you make that’ll end up having AI grow into some unmanageable monster that’ll end up nuking us all and that’s my fear. AI won’t just happen overnight, it’s already here, growing in the background. It’s data footprint becoming ever so much larger each and every second. Let’s look at me for example as a case study (and don’t think you’re any fucking different);
- I wake up in the morning via haptic feedback from my FitBit device (programed to analyse my sleep pattern and wake me up at the optimum time).
- I use voice recognition to get my Amazon Echo to read my emails, last few tweets, switch the Hive lights on in the hallway and play some music whilst I’m eating breakfast.
- At this point I’ve got my smartphone on me, and my smart watch. Both tracking with GPS, WiFi and location services down to within a 2m accuracy. I have these devices on all day, every day. They constantly poll everything from my location to my heart beat and send that back to all kinds of 3rd party servers.
- In my car, my phones Microphone is switched on to allow me to interact with Apple Car Play or Google Maps. Google tracks my location down to pinpoint accuracy.
- I pass 35 ANPR equipped road traffic cameras on the way to work, over 40 private & public CCTV cameras and over 10,000 Wifi access points all handshaking with every electronic device which comes into range.
And the above is before I have my first coffee of the morning. Imagine what this data could be used for? And you think all we’ve got to worry about is some crazy insurgents driving trucks into people in citys?
AI is the single most important advancement in the human race since the Internet, since Penicillin, since mapping the human genome. Get it right and we’ll prolong our existence on this planet for millennia to come. If we fuck it up (yes Government I’m talking to you) by becoming the 2nd most intelligent race on the planet, we’ll be extinct pretty soon through our very own making.
As you were.