How I got deplatformed for my AI work

In the period 2003-2006, I created the basis for AlphaGo and got deplatformed for the sole reason that my opinions irked, and my software financially threatened all major players in the Western Go software/website world.

Mention of my software was forbidden on all major Go websites, including Sensei’s Library. Retailers were told not to resell my software, lest competing vendors would withdraw their stock. Tournaments were warned not to accept my sponsorship. I lost my web host. The American Go Association posted a fake “review” on how shitty my software was. I later discovered their management programmed and sold a competing product called MasterGo. I lost my ability to advertise via Google Adsense, due to a Clickfraud campaign organized for the explicit purpose of delivering the final blow.

Moyo Go Studio

The reason given for my ostracism was that I had excitedly volunteered that I had extracted a small number of non-Copyrighted Go game records from a variety of free- and purchased software. I explained that unless I had access to as many professional game records as possible, it was impossible to do any Machine Learning for Go. Without my efforts and without their replication, the entire field of Machine Learning might have been stalled for a decade and DeepMind would not have come into being. I paid the ultimate price for my work, while DeepMind raked in the dough. At that time, Machine Learning for Go was considered impossible, so people said “You stole those game records to resell them with your useless software, you dirty bastard!”.

Microsoft would later replicate my algorithm after I published it, and Google offered me a job. Tore Graepel, the leader of Microsoft Cambridge’s team, tasked with reverse-engineering my work, later became the team lead at DeepMind, where my work was used to power the plausible-move generator for AlphaGo.

But at the time I was developing and selling my software, all that was directed against me was a seething hatred, borne out of mostly fear and envy. My software threatened all existing Go software, from SGF editors to pattern finders to game databases to actual Go playing programs. Moreover, I wasn’t even a Go player myself! To add insult to injury, I also voiced critical opinions about people’s favorite OSes and programming languages. It rapidly became the consensus that I had to be destroyed – and they succeeded.

The story does not end there, though. I became independently wealthy, with residences in many countries and enough money to not only never have to work again, but also to buy any R&D computer system I could dream of. That’s where the story is now: Being an independent AI scientist yet again – only this time, working on True AI.

As to my future commercial AI software being in danger of being boycotted and my company being deplatformed once more: Impossible. Potential buyers of my software are multi-billion dollar multinationals, Tech giants and even governments such as that of China or the US. Such entities, when they want to buy something, buy it.

128 EPYC cores, 2 TB RAM, 260 TB in 3 x RAID6, dual UPS, tape backup. Weighs 500 lbs, cost $65,000

Disqus Comments