,
11 tweets,
2 min read
Read on Twitter
In the last week I read and listened to 3 articles/news stories all revolving around Artificial Intelligence. All 3 had nothing to do with the military but I believe all 3 have extremely high potential to impact the future of warfare. 1/
The first was about the AlphaStar AI defeating some of the world’s top players of Starcraft 2. The AI learned from watching top player matches to figure out basic strategy and rules, then Google DeepMind (its creator) allowed it to run hundreds of matches against itself 2/
The AI developed a unique strategy which DeepMind then put against 2 top players in best 3 out of 5 matches. The AI continued to learn while playing these players and was able to win 9 of 10 matches between the two players. 2/
Potential Mil Impact: we have catalogued hundreds of simulations in WarSim and recordings from the Star Wars systems at the CTC rotations. We also have a set of rules developed for OPFOR. Why not leverage AI to analyze how we fight and find new TTPs? 4/
Story 2 was about AI needing Data to learn. They call it the unicorn story. They wanted an AI to analyze how we use English to convey information. The data set they pulled was the entirety of reddit with anything that got more than 3 upvotes. ~everything on the internet in 98 5/
They fed the AI 4 sentences to start a science article about unicorns and let it come up with the rest of the story. It was able to analyze the story, find the geographical setting, choose a fake name of a scientist, the university he was from, the region and build a story 6/
The story largely worked except for a weird sentence in the middle about an alien conspiracy (it did learn from Reddit). The military implications are 2-fold: 1) an AI’s concept of reality was shaped by how it procured it’s data. How do we keep an AI analyst from falling 7/
For the same groupthink we already fall into in the military? How do we build objective data? And 2) how long until we see AI build fake news stories in war torn countries further combatting out IO and PA efforts? 8/
Story 3 was about analysts in the Alabama Roy Moore election. Election analysts helped tailor design campaign material to target specific audiences to either get them to vote or not vote in the election. The election was marginally interesting but what I found more interesting 9/
Was the fact that we already use data analysis to target consumers. Mil Impact: An AI set loose to hunt potential extremist/separatist/military youth and convert them for a side. No boots on ground, a war for hearts and minds with no heart or mind reaching out to fight 11/
Final thought: our military is constantly chasing the new shiny. AI will be on the battlefield sooner than we think and if we leave efforts for 5-10 years down the road, then the most technologically advanced country on the planet will lose to cheaper, smarter armies. END/
