Some amazing AI perspectives - that are actually facts
“Soon at a company near you”
Some amazing AI accomplishments
In 2011 an experiment opposed humans to AI over understanding what a blurred image represented, at that time human error rated at 5% while AI at 26%. The experiment was repeated in 2013 and AI error margin dropped to 3%.
Over 2016 AI experiments have been able to:
- “Look” at a picture and produce a short video (out of it) forecasting what would happen afterward. Amazingly it was able to accurately predict things like: “someone falling to the ground”, “someone opening a bottle and drinking”, “a dog running into the water at a beach” and many other.
- Guess what is happening on a given scene just by listening to the audio.
- Predict human behavior on new situations after having analyzed videos of that same human behavior on distinct contexts.
- Generating a picture that accurately shows what someone has described in a text.
These programs were in fact consistently accurate to a 96% degree where expert humans did not go beyond 82%.
An Engineering AI program can create the most efficient airframes for drones just telling it the number of propellers and engine power plus purpose intended for a specific drone. Airbus recently resorted to same technology which produced a cabin partition for the A320 model that is twice as stronger as the original human designed partition yet half the weight.
An AI being released now by “Gamalon” (a Boston Company) can rewrite its own code based on “learned” probabilities instead of hard variables, this alone has the power of making the tedious part of AI development fully automated.
January 2017, AlphaGo beats the best Go player in the world. Why is it relevant? Go is the most difficult game humans have created, it has more permutations in terms of possible moves than all the atoms that have been calculated to exist in the known universe!
By now you will be thinking:
“Well, that’s all very interesting but what does it have to do with Outsourcing?”
In fact, … everything!
Manufacturing – Blue Colar Jobs
In the past, low-cost offshore Outsourcing moved jobs from the U.S. and Europe to other Geographies like India or China. But what does the future hold?
With computing advances and dramatic price drops of hardware components, robots have acquired the dexterity to assemble circuit boards and build cars faster and way more error free than humans.
It is now cheaper to install them on an assembly line than to pay wages to workers in China.
Additionally, while monitoring human’s productivity may be tricky, connected devices and sensors in the Industrial Internet of Things, enable monitoring every aspect of a fully automated manufacturing plant's operation.
China is aware of the threat and is taking the lead in building zero-labour manufacturing plants, with robots doing practically all the work. But it's robots have no advantage over American robots; all work equally hard and consume nothing more than energy. Manufacturing can now start returning to U.S. shores without raising costs.
A fully automated factory that produces cars can also build other products in parallel or cars for different manufacturers, therefore it is now possible to have the Factory Outsourced. In a not distant future, some companies will dedicate themselves to Assembly activities to several clients (like currently other do Logistics).
Another recent example is of AUTODESK that has developed an AI called Bishop which can very well be version 1.0 of a future expert industrial service to be outsourced by companies. Bishop is a precision drilling robotic arm attached to an AI “brain”. The human expert engineer will verbally tell Bishop to drill a hole in a given structure and Bishop will do it, after and if it has analyzed that such hole in that position will not compromise the structure itself. If it does, it will propose an alternative solution for the same purpose.
White Color Jobs
It was estimated that White Color Jobs were safe from being overtaken by Automation in a short horizon, but recent developments have shown otherwise.
As we have seen in recent years AI has evolved exponentially and such software is now capable of solving some complex challenges faster than humans and with low error margins so Virtual Secretaries that assure you both prompt assertive e-mail answering and a schedule which obey by your preferences and prioritizes your most time profitable meetings is already in the market. Maybe not capable of replacing humans, but who knows within a 2 to 5-year time timeframe?
We also assessed that breakthroughs in AI lead to the creation of diagnosis software that is more accurate than Human evaluation.
It was found that two ophthalmologists (eye doctors) would only fully agree on a diagnosis 60% of the times, well, fortunately, AI has proven this year of 2017 to be 97% accurate on such diagnosis.
Some CT Scan diagnosis software can detect in average 20% more potential health threats in the exam than the “human eye”. One example of a cancer probability CT scan software has recently diagnosed 10% more high potential cancer developing features on cell clusters in several patient’s lungs film tests.
Nowadays we can have a powerful CPU and some nice amount of RAM literally on our wrist, on a Smartwatch which also has some embed sensors that can permanently monitor our balance, blood pressure, heart beat and temperature. Having those data forward to an online health platform once every hour and a local app on the wristwatch that triggers an alert if some strange pattern is detected can save lives. An AI based health platform may leverage information gathered over time from a patient and allow not only accurate medical track record that a physician can analyze as also advice on adequate treatment options based on detailed organic behavior compared to that patient’s medical history (gathered over time).
All the herein mentioned services can be Outsourced by health professionals as other businesses.
Customer support AI automation has in fact already begun some 5 years ago. It is still rudimentary AI in most cases, but some “assistants” to whom you speak when calling a support line are in fact AI programs that will lead you through several steps in a “personalized” friendly manner to “understand” your need.
In a near future you will have a “conversation alike” interaction with such software and in most cases not even noticing that there is “no one” on the other side of the line.
History has proven that automation has done more to develop and reshape human jobs (by means of assuming heavy and repetitive tasks) than really to still those jobs. Two quick examples:
- Robots in dangerous and critical environments on Industry grounds have ended up saving lives and improving staff qualifications over time.
- ATM machines did not make all the cashiers lose their jobs, in fact, they have allowed the economy to grow by raising currency flow which leads to banks opening more agencies and shifting the traditional cashier work to more complex and qualified client facing services (including investment coaching that further raises the bank’ profits).
The AI revolution will severely impact society, even if it is a step by step process. It is inevitable that some jobs will be lost and to avoid the affected individuals to become economically irrelevant (slipping to the outskirts of the social pattern), governments need to think SMART and prepare programs that will requalify those persons, making them an added value in this new era instead of a social problem.
Everybody can learn anything,
it is a matter of time and motivation
From all the theories and fears one thing emerges as a logical conclusion:
Once humans reach the stage of achieving AI development that overwhelms their rational capacity that will be the very last thing humans have created. From then onwards AI will be doing the creative thinking, so what about humans?