The field of Artificial Intelligence is blooming at the moment, with companies like Google and Alexa delivering small, inexpensive devices to your living room for $75 that can listen to you talk and tell you the weather, order pizza and turn your TV on.
However, a Christchurch doctor and a University of Otago professor have been trialling an amazing new AI (opens new window) that, at first glance, sounds like it's a huge leap ahead of the competition. Whereas Google's Home and Amazon's Alexa both understand simple spoken instructions, Zach from the Terrible Foundation understands natural language (how we normally speak) and reads text in the form of emails that are written in a conversational style.
This week, David Farrier has written a couple of articles about the AI (opens new window) and the organisation "running" it (opens new window). David is known, amongst other things, for his previous work writing about Scientology and uncovering the seedy topic of competitive tickling, which he chronicled in his documentary Tickled.
Given that it was David writing this article, it was likely that there was going to be something interesting about this AI. David laid out a bit about the history and organisation of the company - Terrible. I've done some digging of my own, and follow up articles have been written, showing that what looks like a large, philanthropic organisation - the Terrible Foundation, is likely just a young man and his father - Albi and David Whale. The organisation lists many branches, but appears to have copied a lot of its text from other websites such as Oxfam.
Speaking of Oxfam, the organisation is registered as a charity (opens new window) and says that it hands all its profits to other charities, such as Oxfam. However Oxfam said they broke their arrangement with Terrible several years ago, and that not much money was received. Charities records support the fact that not much money has been given out. Weirdly, the records also show that the Terrible group reported gaining over $400 million in assets last year - which apparently is the donation to the organisation of the hardware that runs their AI.
On to the AI. it's called Zach, and it's apparently made from custom silicon and is Nitrogen cooled. Zach supposedly took over the running of Terrible the organisation, doing everything from accounts to helpdesk.
A doctor in Christchurch has recently been testing Zach by sending it audio recordings of anonymised patient consultations, and receiving typed up patient records. The responses typically take about 20 minutes, which David points out in his original article as being a red flag.
The next big red flag is that Zach, the AI, seems to make frequent spelling mistakes. Funnily enough, so does Albi, the man who is behind the Terrible foundation. His press releases and forum posts online seem littered with typos. It's been explained away as a layer of software called the Enigma layer, where the computer tries to mimic the person talking to it - but to me that seems to just be lies piled on top of more lies. It's nonsense to argue that people who make mistakes in their writing also prefer to read text that has mistakes in it.
I can't help but think, when reading through historical articles about Albi, flicking through old press releases for his companies and seeing his rebuttal to David Farrier's articles, that Albi just can't help himself. For every inconsistency, there's an implausible but not impossible excuse - and from where I stand, it looks like Albi just needs to stop digging and put the spade down. I imagine that, if indeed Zach does not exist, Albi will already be in a lot of trouble for the promises he's made and the people he's pulled into his world. It would be best for him to think of ways in which he might be hurting other people, and stop the madness before he ruins people's lives. Known scam artists in the past have destroyed countless lives with their smooth talking and big promises.
There's an IPO coming up soon, and David's shared a prospectus on twitter. There's nothing of use in there to be able to tell if Zach is real. The best advice at the moment is to not invest until the Terrible Foundation has proven beyond a shadow of a doubt that their supercomputer exists and can do everything they say can, and that "Zach" isn't just Albi answering emails and pretending to be a computer.