adversarial examples – Tesla tricked to change lanes in broad daylight

Tiny red stickers allegedly can be used to confuse Tesla’s understanding of road markings. As a result this group claims to be able to make Tesla change lanes in broad daylight no snow or rain needed. https://keenlab.tencent.com/en/whitepapers/Experimental_Security_Research_of_Tesla_Autopilot.pdf

If autonomous cars are really smart , they ‘ll keep moving to slash parking costs

If autonomous cars are really smart , they 'll keep moving  to slash parking costs
Well Given very low costs per mile driven and high parking fees this seems to be the optimum right? To add insult to injury, it makes perfect sense to cruise at nearly zero speed, to expend least amount of energy. So, how do you like the idea of cities crowded with autonomous cars driving around ...

Amazon gives up AI recruitment tool – it disliked women

I would love to have a look inside their model to see how it was set-up. Apparently the rating tool was not appreciating women. Simple classification, clustering task. Decision Trees? Neural Networks? Too simple or too complicated? Probably they would never disclose the details. YAFU. https://tech.slashdot.org/story/18/10/10/1547229/amazon-scraps-secret-ai-recruiting-tool-that-showed-bias-against-women

I Love You. Over 10% of Gmail replies comes from Smart Reply

As you may have read here: https://www.businessinsider.com/google-changed-gmail-smart-reply-after-ai-kept-suggesting-i-love-you-response-2018-9?IR=T Google’s AI tool scans emails, headers, responses and the like to pursue ideally tailored responses custom generated for you. At Tbe begenning, the favourite ones though were: “I Love you” , ‘Sent from my IPhone”.  Possibly neural network output. Isn’t that predictably cool coincidence?

Communicators Security Issues Summary

Source http://niebezpiecznik.pl Enjoy your WhatsUp 🙂

Amazon’s Facial Recognition Wrongly Identifies 28 Lawmakers, A.C.L.U. Says

1
Dzień Dobry
Jak możemy pomóc ?
Powered by