Defined, Complex Systems Will Get Automated by AI, This Will Be 80% of Law and Finance Jobs

Vivienne Ming is a theoretical neuroscientist, technologist, and entrepreneur. She co-founded Socos, where machine learning and cognitive neuroscience are combined to maximize students’ life outcomes. She sits on the boards of StartOut, The Palm Center, Emozia, Engender, and Genderis Inc., and is a Chief Science Advisor to Cornerstone Capital, Platypus Institute, Shiftgig, and Bayes Impact.

Vivienne has developed a predictive model of diabetes to better manage the glucose levels of her diabetic son and systems to predict manic episodes in bipolar sufferers.

Vivienne defines artificial intelligence as an immensely powerful tool. AI is not the same old thing.

AI is any autonomous system that can make decisions under uncertainty.

– how much financial risk is there
– what are the objects in these pictures

AlphaGO and AlphaZero that are better at playing GO than any human player, do not actually understand GO. The AI image recognition systems do not understand images. This can be seen in the errors that they make.

The AI game systems can play against each other and gain 125 years of experience in a day. They can explore all the aspects and approaches to different games.

Low skill jobs are actually very hard to automate. Picking strawberries is actually very hard. Transforming the farm to raised hydroponics and having the strawberries handing down makes it easier to automate.

80% of the financial jobs will be automatable by AI systems. The financial worker and legal worker jobs have high pay which make them good targets for AI.

In a competition at Columbia University between human lawyers and AI legal systems to work on analyzing legal contracts to look for loopholes. They used NDAs (Non-disclosure agreements) with engineered loopholes. The AI’s found 95% of the loopholes and humans found 88%. The humans took 15 minutes to read each contract and the AI’s took 20 seconds.

AI Deep Learning can generate and amplify experience, particularly in defined systems. This is the case even when the experience and the rules are complex.

Vivienne predicts that the next wave of AI will be naturalistic, closed loop and causal. The super-human performance is currently only in defined domains. The three new capabilities will open AI up to areas with more real-world problems and less defined areas.

Google is adding graphical networks to provide AI with representations of what we already know to be true.

The naturalistic systems would teach AI using sensors and data observations of the real world and not probably biased data sets.

Vivienne thinks AGI is very far away and requires new approaches.

Vivenne thinks we will soon have AI augmented human intelligence. This will enhance people in general ways. Systems will enhance your working memory by 15%.

We can choose to improve and enhance humanity.