AI's cloudy sky

2018. October 11
A munkatárs fényképe
associate professor
Szoba: IE437
+36 1 463-2899
Email: meszaros (*) mit * bme * hu

Az előadás tartalma

AI technologies are easier to use than ever thanks to cloud computing but can we deal with the problems on the rise like machine bias, missing explanations and data quality in machine learning, technology misuse etc. This is a (very) shortened version of my previous presentation in Hungarian.

Az előadás anyagai

Reading list

AI and  Cloud Computing

MLaaS: Google tutorial
Google ML Engine, Microsoft Azure Cognitive Services and Machine Learning Studio, Amazon Machine Learning, IBM Watson Analytics
and (Stanford)
Ars technica
Cloud GPU


Google, Facebook, Amazon Alexa, Microsoft LUIS, IBM Watson Conversation and its applications, Oracle bots,
Chatbot customer services
Cloud example
They can lie.

Data quality issues

AI in Risk assesment and race bias, paper
Image processing in self-driving cars, Uber driver and passenger hack, Tesla accident,


brain imaging
"life mining", mass surveillance, intro, citizen score in China, "karma police" in the UK, NSA PRISM in the US.

Can we understand how it works?

Nvidia self-driving car and a paper about it
Nature article

Military AI

Slaughterbots and protest against it (Stuart Russell was  our guest in 2018)
Newspaper article

AI safety

Introduction and videos, TED-talk
Max Tegmark (MIT, book), Yann LeCun (Facebook AI), Elon Musk (Tesla), Bill Gates (Microsoft), SciAm
Researchers and their call.
A study for the US gov
Value alignment: Stuart Russell (Berkeley)


Self-training AI
Hardware: ARM, Huawei, Apple,

© 2010-2024 BME MIT