Published on October 12, 2017
Artificial Intelligence (AI) – The must-digestive technology concept of IT professionals and the buzzword for rest of the world and was born before 60 years!! Sounds buzzing right!! A proposal for the Dartmouth Summer Research Project was submitted on 31st August’ 1955 by American Mathematician and then a Computer Scientist J. McCarthy – Dartmouth College and his team of 4 members. McCarthy was credited for coining the phrase ‘Artificial Intelligence’ many years before our birth!! Later in July’ 2006, The Next 50 years (AI@50) was held on 50th anniversary of AI’s birth to assess where AI had progressed in 50 years. 5 of the 10 original attendees including McCarthy attended AI@50 and took AI vision further and discussed projections to 2056. Of-course, previously defined benchmarks of AI are outdated and revised rapidly in last 10 years but the journey of initiation has been too long…
That was high level history of how AI was originated.
Next – What is AI / MI?
AI or MI is intelligent computer programs processed by machines (computer systems or robots) that exhibit same intelligence as humans. These programs learn by rules, self-learning and self-correction with lots of data analytics.
AI is Not One Technology; or rather Not Only Technology.
AI is mixture of multiple technologies like Machine Learning, Computer Vision, Deep Learning, Natural Language Processing (NLP), Neural Network, Robots, IoT, BigData and others inspired by biological, mechanical, biomedical systems and finding solutions of important human or humankind problems.
A not very old example of AI was the Chess championship in 1996 and 1997 @ Philly and NYC respectively between the world champion Garry Kasparov and IBM super computer Deep Blue where the man was defeated by the machine! AI programming in video games at gaming zones are another older example of AI in practice. Autonomous and Self-Driving cars is another very popular example of AI which we all are aware of.
So, I want to talk about few unusual and interesting use cases of AI here.
1. Predictive Machinery Maintenance
For high-end plants and machineries and their parts like batteries, gear-boxes, air conditioners and different controller units, AI has started helping rapidly in terms of identifying failures before it occurs.
AI based algorithms are developed by integrating robots, sensors, switches (IoT devices) etc. for capturing sounds, vibration specific data continuously. Pre-defined data set is already made available for correct and faulty signals and with large set of data being collected, this set is made self-matured with smart algorithms. With differences of sounds and vibrations detected; predictive analysis of breakdown is shared at larger level for initiative preventive actions further.
German industrial sectors have been putting remarkable result oriented efforts in this area.
2. Super Smart Society Concept
Japan has been leading in AI since beginning (but yes, undoubtedly, Silicon Valley is picking up very fast in this area recently). By leveraging AI, Robots, IOT and Big Data; Japan Government has launched the plan of setting up the foundation for a smart society by 2021. The use case described in Japantimes mentions this with this example: When a man asks his robot for a suggestion on what he should eat for dinner; the robot will suggest the food matching with his health condition – as per health data automatically retrieved by the robot through health device being worn by the person – will get confirmation by man through interactive voice communication – will order the food from farms through drone – after the drone delivers the food – robot will install agreed recipe data to a cooking machine – and prepare dinner for the man!
Such and similar concepts are being contemplated by Japan Government as Super Smart Society, where all individuals can receive the necessary services they need regardless of age, religion and language factors.
3. Improve Data Centers Efficiency
DeepMind, originally into AI based games (inventors of AlphaGo, online Chess program) was established in London in 2010 and acquired by Google in 2014; is working for improving energy efficiency for Google Data Centers. The system collects energy consumption data every 30 seconds and by applying concepts of machine learning and neural networks to optimize data center operations by consuming low energy at the right time. Data like total IT load, outside and internal temperature are measured continuously, run it through a model that find hidden interactions and provide recommendations and implement optimizations recommendations to effectively use energy to run data centers. Usage of water for e.g. instead of air conditioners to cool data centers is also applied for taking care of outside world temperature. As per DeepMind report, using AI technology, energy efficiency increase has been achieved up to 40%!
4. Analyze Crime Patterns
The concept of predictive policing is based on constructing M.O. (modus operandi) of criminals. Based on years of data collected about crime conducted by individuals or groups, the behavior and approach of crime is provided to police (E.g. time of crime, back door / front door / window entry, type of property chosen and many more). This helps the department to right away start investigations based on data available. This machine learning method can be a very effective tool for crime pattern detection and help police to speed up their operations rather than waiting for weeks and months to discover patterns.
5. Deal with Cyber Security
Machine learning programs are being developed on various algorithms to define defense approaches.
One such approach is after constantly analyzing and learning about correct data, false data can be inserted through ML technique to prove attackers wrong in classifying right data.
Major first generation AI based security solutions are focused on data sifting, risk identification and mitigation and auto-remediation of known threats.
But at the same time the question comes to my mind is that one level higher techniques may be developed by hackers as well based on same ML techniques to break defender’s techniques!!!
6. AI Gearing-Up AgTech Domain
With the challenges like need of serving 7.6 billion people around the world which is targeted to 8 billion by 2050, increasing infrastructure subtracting farming lands gradually and farmers decreasing gradually, efficient use of technology is essential. With this emerging agriculture technology (AgTech) companies, start-ups and companies funded by giants like Google, Amazon and Microsoft are researching a lot in AI to tackle this challenge and agriculture.
As an example, a technology use-case is developed that uses machine vision to identify ripe apples and then apply vacuum mechanism to pull those apples from the branch automatically.
Another example is of image recognition algorithm development to detect and classify plant diseases like fungal and others, pests and either pluck that portion or precisely spray specific medicine on that part of the crop.
There are other scenarios also where data of soil sampling, weather effects etc. are being gathered and gradually AI use-cases are being built upon that for various agricultural benefits. Yes, this process is relatively slow as data can be captured only once or twice a year depending on the time required for crops to grow completely but future will surely be benefited by AI in this industry.
7. Understanding Movie Line-up by Movie Making Companies
The Walt Disney Co. who has given multiple box office hits as The Beauty and the Beast, Star Wars, Captain America: Civil war and many more… has started looking to understand their movie-audience better through facial variational autoencoders built on AI concept. The analysis jointly done with NVIDIA is based audience facial expressions, how much eyes were open, smile and laughter etc. through IR cameras in cinema-hall. A 10-minute analysis on each person gives judgement of individual’s choice and how s/he will perceive the movie. Movie edits can be done based on this data for from money making, success and crowd pleasure perspective.
8. AI is a blessing for Healthcare!
Healthcare organizations are becoming much IT savvy gradually. As the data shows, healthcare firms were investing 4.2 % revenue in IT in 2014, which increased to 40% in 2012 and now it is at 67% this year! Big companies like IBM, Google, Microsoft and Intel are working with healthcare giants like CVS health, Johnson & Johnson, Oregon Health and Science University, UK National Health Service etc. Positive results are being derived in generating precautionary signals for various diseases, oncology research areas, wellness segments and many more. Medi-Tech startup’s contribution in this area is also at huge scale.
A funded company Proteus Digital Health in Silicon Valley has invented a HIPPA, FDA compliant digital pill that captures various health specific data, keeps track of medication prescribes by doctors and its impact in the body. The data is transferred to mobile app using Bluetooth and are transmitted to doctors through internet. From data analysis; predictions and alarms are generated and changes in treatment are done accordingly.
Machine learning technology uses large amount of trend analysis data to predict alarming stage of the disease at early stage. Such analysis with more than 52% accuracy is already being done for detecting breast cancers in women before officially diagnosed by medical reports. One more such example is a company using algorithm that identifies eight variables which predict avoidable hospitalization for diabetic patients based on related health data trends.
A solution developed in Japan for diagnosing a rare type of leukemia is a remarkable contribution in saving a patient’s life as conventional tests had failed to detect the disease.
There are examples in areas of AI being used for precise medical insurance and other sectors.
9. Coco Cola using AI for Inventing New Flavours!
World’s largest beverage company Coca Cola having its presence in all 195 countries, having 500 brands and serving the world 1.9 billing servings every day is the first non-IT brand to talk about AI and Big Data.
Vending machines of their new brand allows customers to add drinks of their choices and customize the drink. Coca Cola’s plan is to pick the most popular combinations and launch as a ready-made canned drink.
Another new launch of Coca Cola is AI bots (Siri and Alexa) in vending machines. These interactive programs facilitate users for mixing flavors to their preferences and many more. Machines also learn based on location specific behavioral changes. This is just the beginning.
Concepts are sowed and branches to extend unconditionally.
10. Recruitment Pointers
AI is helping HR and recruiters in faster and higher authentic candidates search rapidly.
AI based video intelligence systems analyze video interviews of candidates and provide insights to employers.
SaaS based AI platform browse social profiles using predictive algorithms to determine which candidates were in mindset to leave their current jobs.
Interactive automated screening tests for programming, personality and language assessments are also widely used by large number of Fortune 500 companies with highly satisfactory results.
Specially there is a surge in valley of start-ups working in this area and strongly supporting big giants in skills recruitments. Looking at the trend, it looks very clear that existing job portals will be heavily impacted by services of these start-ups in this area.
Several open source and proprietary tools and frameworks are available e.g. tpot (Python), auto-sklern, Auto-WEKA, machineJS, Watson (IBM), Myraid X and NCS (Neural Compute Stick) (Intel and Movidis) and many more for developing artificial intelligent solutions.
The power of AI lies is providing structure or enabling horizons to find important data automatically and utilise this data for improvements and achievements.
Over the last decade over $17b has been invested in various AI related start-ups in the US itself. Japan is already A Pioneer in AI inventions and China is also taking larger and quicker steps.
There are right questions around hacking, security, job risks etc. but as per the definition; AI is constantly learning. Learning in all directions where improvements are required and so the future is very promising and positive.
And experts correctly say: Everything that can be automated, will be automated!
So, there is no option for all of us but to digest the nature of AI: Capture a lot of data, learn from this data and use learnings for Better Us and Better World!
(Image credit: pixaboy.com)
(References from Forbes.com, Wired.com, Japantimes.co.jp, mckinsey.com)
