AI in Industries

AI is becoming a live-or-die decision for many companies. At the same time, it provides vast opportunities for companies who can leverage it quickly in their business. If you want to turn your industry expertise and data quickly into some cutting-edge AI solutions, either help you gain competitive edges over the competition or create new business opportunities, R2.ai is here to help.     Get in touch now
Get in touch now
Here are some examples of how various industries are adopting and exploring AI technologies, and hopefully they will help create some new sparks of innovative ideas of your own.
Select an Industry:
Select another Industry...

Healthcare

In the healthcare industry, machine learning and artificial intelligence have quickly rocketed to the top of the industry’s buzzword list. We have observed the following key trends.
Preventative Care
Preventative care/treatment is a set of services designed for disease prevention before the disease happens or before it gets worse. It can help people stay healthy and live longer. It also helps make the health providers more efficient and cost-effective. At the core of an effective preventative care system is the ability to accurately predict the disease risk based on data points such as environmental factors, genetic predisposition, disease agents, and lifestyle choices. Machine learning and AI plays a vital role in building the predictive models.

Here are some interesting examples:
• Predict chronicle conditions and disease progression risks
Using predictive models built on classic machine learning or AI system, healthcare providers can use characteristics from current patients to predict patients at risk to develop certain chronic conditions, which allows for earlier intervention to reduce the cost of care and provide better care for the patients.
• Predict conditions that could result in hospitalization
Using electronic medical record data in machine learning/AI predictive models to estimate the likelihood of hospitalization, healthcare providers can then effectively allocate resources to proactive intervention to prevent potential hospitalization.
• Predict need for rapid response team
Using data from the electronic medical record to predict which patients are at highest risk for needing an intervention from a rapid response team, also known as a medical emergency team. Healthcare providers could use this model to proactively position rapid response team members and equipment to decrease the response time.
• Conduct cause of diseases analysis
Combining domain expertise and machine learning to identify the key cause of certain diseases.
Computer-aided diagnosis
Computer-aided
diagnosis
Improving imaging analytics and pathology with machine learning is of particular interest to healthcare organizations, who has vast amount patient data available. Machine learning can supplement the skills of human specialist by identifying subtler changes in imaging scans more quickly and help provide earlier and more accurate diagnoses.

In one research, machine learning based models have already exceeded the accuracy of human pathologists examining images of metastasized breast cancer tissue, reducing false negatives to one-quarter of the human clinical rate. In another research, machine learning tools performed better than human pathologists when distinguishing between two types of lung cancer.  The computer also beat its human counterparts at predicting patient survival times.
Precision medicine
Precision Medicine is about tailoring of medical treatment to the individual characteristics of each patient. Using precision medicine for cancer treatment as an example, it is important to identify a specific treatment’s effectiveness for a specific patient. Preventive or therapeutic interventions can then be concentrated on those who will benefit, sparing expense and side effects for those who will not.
Medical procedure risk analysis
Medical procedure
risk analysis
Predicting the risk of a specific patient treated with a particular medical procedure is fundamental to medical practice. Machine learning and AI technology can tremendously help improve medical procedure decision making, based on the patients’ test data, medical history, and personal health data.
AI-based surgeon training
For surgical training, AI Technology can be used to analyze images to identify best practices from top surgeons that could be fed back into simulations to improve surgeon training over time.
Robot surgeon
Robot surgeon technology today enables surgeons to perform complex operations with good precision. New AI research now tries to make the robots more intelligent. One of the projects is to use artificial intelligence to help surgeons interpret what they see inside a patient. Existing robots just leave surgeons to look at a video feed. The artificial neural networks used for image search could annotate a feed with anatomical data and guidance such as information about the boundaries of a tumor. In this case, AI technology could give the surgeon a level of expertise usually obtainable only after experiencing thousands of cases.

Insurance

The challenges for the insurance industry have always been heavy data handling and complex business process. What is more, even the traditional product lines such as auto insurance may shrink by a whopping 60% over the next 25 years, when “radically safer” vehicles including the driverless technology are in used, according to KPMG report.

With these challenges, the insurance industry is perfectly poised to adopt AI technology and machine learning, to streamline operations, improve services, and create better insurance policies that enlarge the industry landscape.

Here are some examples how AI technology and machine learning can help.
Insurance advice
Who doesn’t want personalized insurance advice with no wait time and no unsolicited advice? In a recent survey, nearly half of the consumers are already open to robot advisor, for it’s more convenient, accurate, and faster.

With robot advisor empowered by AI and machine learning, customers can now get easy access to recommendations that are most relevant based on their profiles, tolerate level, and real-time market data, in no time. Meanwhile, the time it could save insurance agents by not having to answer basic questions and handle routine admin tasks will free them up to concentrate on more complex requests.
Claim processing
Handling the claims is often time such a hassle for customers. The company’s reputation to process claims is an important criterion when it comes to choosing insurance providers.

AI technology and machine learning can dramatically improve the claim processing from initiating a report to moving claim through the system, to communicating with customers and to eventually making the payments, with little to no human interaction. Think about it, what if, customers can get their claims paid in minutes? The results are higher customer satisfaction, lower costs, and more business for the company.
Fraud prevention
Fraudulent claims are costing insurance companies a lot of money, not to mention the costs to carry out a massive investigation.

AI technology and machine learning can help to identify those high probability frauds better and faster than human, based on historical data on fraudulent claims, real-time data of customer behavior, and customer profile. From here, a further human investigation can be conducted for more targeted potential frauds. And when frauds are confirmed, the data can be fed into the machine learning algorithms to further “learn” and improve the model. In this case, human and machine can work better together than human alone.
Policies development
Imagine, with a vast amount of data at your disposal, if you could use it to the full potential, you would be able to create policies that serve your clients better, and manage your risks better.
• Usage-based insurance
In the car insurance market, for example, usage-based insurance allows customers with better behavior and healthier lifestyles to pay less. In this case, IoT sensors can feed personal data, such as driving data, driver activities, and heart rate, to the machine learning algorithms, to identify and classify drivers’ behavior and health situation; based on which, the algorithms decide what types of policies and pricing shall be applied for the customer. This is radically different than the traditional approach because machine learning enables policy and pricing based on the actual behavior of individuals rather than likely behavior based on categories.
• On-demand insurance
On-demand insurance is another example, where AI can pull customers’ personal data, financial data, geographic and social data to customize insurance coverage for the specific event. In this case, machine learning enables policy and pricing based on real-time data linked to actual individuals, rather than historical data linked to statistical sample groups, to make the best prediction.
Reserves building
The insurance industry will always involve risks. Building reserve is critical for the business to mitigate risks, allocate funding, and improve margins.

Machine learning can be well used here to build business models that can automatically process and analyze data for more accurate results, based on past losses, predictive data on future losses, competitive information, and real-time market data, so that the company can manage the reserve more systematically and accurately.
Marketing Strategy
Predictive analytics is very useful to set marketing strategy as individualized as possible, for customer acquisition, retention, and products cross-sale.

In the past, the analytics is primarily descriptive and historical. Today, AI technology and predictive analytics make it possible to build business models that can integrate and analyze not only historical sales and marketing data but also real-time data, such as social media, behavior data, what customers view on websites, etc., to offer high confidence in the prediction. This leads to better business insights that can form a complete customer profile, which will greatly improve marketing strategy.

Automotive

The competition in the automotive industry continues to heat up. Innovative technology, globalization, shifting market conditions, and heightened customer expectations are leading to a changing landscape in the market.

AI technology is becoming the key driving factor in innovating the automobile industry. It can be so widely used literally in all major subprocesses in the value chain, from development, procurement, production, marketing, sales, to services.

In 2015, the install rate of AI-based systems in new vehicles was just 8%; according to a recent forecast, this number is expected to soar to 100% in 2025. Here are a few use cases to help you re-imagine the revolutionary possibilities that AI technology can offer.
Development
There are virtually unlimited potentials for AI in car development. Here are a few examples:
• Driverless car
So who drives the driverless car? The answer is artificial intelligence. The beauty of devices with AI is that it tries to learn from sensory inputs like real sounds and images, and AI would recognize the environment and evaluate the contextual implications. Companies like Google, Tesla, Ford Motor, and many others are spending millions of dollars in research to come up with better technology and to make driverless cars a commercial reality.
• Autonomous system
Even if you still want to drive a car, AI technology will put data in the driving seat that help you be safer, more informed, and more enjoyable. With the help of AI, vehicles can be autonomous and smart. For example, machine learning can automatically identify and classify drivers’ behavior, create user-specific profiles, provide a user-based setting, to enhance user experience. Some companies have gone a step further to create highly responsive autonomous systems that aid in mobility for those who are "less able to drive." The AI-powered smart car can make decisions for you the way you like it, and execute.
• Cognitive capabilities
Machine learning along with image/voice recognition can add cognitive capabilities to interpret a real-life situation and understand unstructured data. Vehicles can learn the environment and adapt accordingly, by analyzing real-time data such as traffic information, nearby cars’ driving patterns, road situation, weather forecast, and destination updates. Today, automobile companies are adding cognitive capabilities to its cars to help vehicles to handle dynamic conditions on the road.
• Connected car
AI-empowered connected car can be a social agent, it can “contribute” traffic information to improve travel time, “communicate” with other cars to avoid accidents, “speak” with manufacturers to diagnose machine problems, and “identify” drivers who enjoy similar car driving experience.
Procurement
Procurement uses a wide variety of data that is large volume and extremely tedious. With the heavy data and timely decision-making requirements, machine learning and predictive analytics can do a better job to provide business insights to lower costs and eliminate bottlenecks than human. To optimize stock level, for example, a machine learning business model can pull data from the model sold, model age, sales forecast, service records, real-time car condition data, inventory, supplier data, to generate a more accurate purchase forecast promptly than a human can do.
Production
The production process is complex and needs to be precise to high-quality standards. Today many productions are still experience-based. To translate the expertise and data to achieve defect-free manufacturing goal, machine learning based on reading and analyzing the vast amount of data, structured and unstructured, is critical to improving production.
• Industry robot
Industry robot that makes and assembles cars cannot just operate according to statistic programming; it shall also be able to use machine learning to work autonomously towards defined learning goals. For example, when an issue is identified through car service, the data can be traced back to production, to analyze the possible causes of the issue, what is attributable to production, and set the learning goal for the corrective measure. In this case, machine learning systems can “see” much better than humans, and require no human intervention to provide more accurate analysis and solution promptly.
• Predictive diagnostics/Preventive maintenance
Today, automobile manufacturers strike to make telematics and connected car work to provide predictive diagnose and preventive maintenance, which can improve car quality, security, and reliability. According to one global car manufacturer, the project can provide savings of up to $800 per car. The key in these savings lie in connected cars that can communicate with the manufacturer via 4G or LTE for data provided by sensors that measure all moving parts of a car; the manufacturer can then provide subscription-based in-vehicle remote diagnostics. These real-time diagnostics can check the health of the most important systems within cars, send it back to the carmakers, and use the data to build maintenance preventative models, or use the data to improve future car design.
Sales and Marketing
To stay ahead in a saturated and competitive market, it is critical to set the right sales and marketing strategies that are as individualized as possible.

Many car manufacturers have not yet taken full advantage that predictive analytics can offer, and not having full control over various data sources especially external data such as social media to gain insights.

Predictive analytics is imperative for companies to analyze a variety of data, external and internal, structured and unstructured, and at the individual level. For example, the analytics can gain insights on which customers may like the luxury or economic cars, where are they located, when they may consider buying, what’s their financial and social profile, and what may be the optimal offer. The more you know about your customers on an individual level, the better you could address their needs, and pitch the sale through the most effective channel.
Services
With AI technology and predictive analytics, many car owners do not need to worry about when and where to schedule their car services anymore.

For car owners, the system will automatically check car usage data, monitor machine condition, predict what part is likely to fail, remind drivers for car service, and book the service appointment with preferred repair shop nearby. And if a car service is scheduled, performance data will be sent to the repair shop and carmakers --- AI technology and predictive analytics can do it all for you!

On the other hand, for car companies, the repair shop service data can be well used for predictive analytics, to predict what may trigger a warranty service or recall, what is attributable to production, in order to improve car quality, forecast warranty costs, and even identify potential sales opportunities.

Financial

Given the high volume, accurate historical records, and quantitative nature of the financial service world, few industries are better suited for artificial intelligence.

Resisting machine learning and other AI applications won't stop their proliferation because the potential benefits are just too great. McKinsey advises executives that "now is the time to grapple with these changes because the competitive significance of business models turbocharged by machine learning is poised to surge.”

Today, AI technology and machine learning have come to play an integral role in many phases of the financial ecosystem, from approving loans, to managing assets, to assessing risks, and to improving customer services.
Financial advisory
According to Tech Emergence, an AI market research firm, machine learning can provide more personalized and calibrated applications to advise on financial products, it may be perceived as more trustworthy, objective, and reliable than human advisors, just as Amazon and Netflix can recommend books and movies better than the human expert.
• Robo-advisor
Robo-advisors have gained significant traction with consumers in the last couple years especially with millennials, who are less willing to pay the fees to in-person advisors. Robo-advisors can develop algorithms to calibrate financial portfolio to the goals set by users. Users enter their financial goals along with personal information such as age, income, and financial assets; the system then automatically spread investments across financial instruments based on real-time changes in the market, always aiming to find the best fit for the user’s goal.
• Digital financial assistant
Banks are making a bold push into the AI-empowered digital financial assistant, leveraging predictive analytics and cognitive messaging, to provide financial guidance to their customers. The digital financial assistant offers customers a solution to help manage their basic banking needs more efficiently and consistently, 24/7. It is available through voice/message chat on the bank’s mobile app, which enables you to make command, for example, send money to a friend or pay a bill; it also leverages predictive analytics to help you achieve a savings goal by making suggestions based on your income and spending patterns. Meanwhile, this solution allows the human financial advisors to focus more on supporting the customers’ more complex needs.
Trading
AI systems enable hedge funds and financial institutions to make high-frequency trading (HFT), as millions of trades are made in a day. Machine learning is playing an increasingly important role in calibrating trading decisions in real time.
• Robotic process automation (RPA)
Robotic process automation (RPA) integrates AI and is carried out by web robots programmed to process repetitive tasks, to improve operational efficiency and reduce costs. One bank that uses RPA reports that the implementation has led to the following results:
o 100 percent accuracy in account-closure validations across systems
o 88 percent improvement in processing time
o 66 percent improvement in trade entry turnaround time
o ¼-second robotic reconciliation of a failed trade vs. 5-10 minutes by a human
RPA also help to reduce costs significantly. The bank estimates that the “funds transfer bots” alone contributes to $300,000 in annual savings, by cutting down on the time employees spend processing payments and resolving data errors.

Automatic stock trading
An AI company that also runs a hedge fund has developed an algorithm that ingests millions of data points to find trading patterns and forecast trends, which enable it to make successful stock trading decisions. The system runs trillions of simulated trading scenarios created from the vast amounts of public data available online. These techniques enable the startup to squeeze 1,800 days of trading into a few minutes. It's algorithms use those scenarios to identify and blend successful trading patterns and devise new strategies. Successful trading strategies are then tested in live trading, where they evolve autonomously as they gain experience.
Foreign currency trading
Some financial institutions are already developing AI FX trading systems. One organization uses AI to trade four major currencies says that AI keeps trading costs down, and has generated $100 million from an investment of $10,000 in just two years, a result that seems almost too good to be true. The advantages of AI-empowered machine learning are that it can take into consideration of many micro and macro economic variables that influences currency market, and are objective, self-learning, and flexible.
Fraud detection
When increasing amount of valuable company data being stored online, the data security risk also increases significantly. Previous fraud detection systems depended heavily on complex sets of rules; modern fraud detection goes beyond to use machine learning to actively detect certain activities that may be potential security threats and flag them for security teams. When the frauds are investigated, the results can be fed back into the system, so it can “learn” and further build the customer profile, and enhance risk detection model.
• Credit Card fraud prevention
Credit card companies have started to use the billions of transactions it processes every year to train machine learning algorithms. It gleans patterns from historical spending habits of cardholders to set a behavioral baseline, against which it will compare and score each new transaction for each cardholder. This is a major improvement over traditional prevention technologies based on predefined rules that rely on a one-size-fits-all approach to evaluate all transactions.
• Real-time fraud alert
Machine learning algorithms work in real-time to identify and eradicate fraud including online and in-person banking. It utilizes machine learning to continuously and rapidly evaluate the correlation of customer data from multiple different sources and potentially fraudulent activities. Fraudulent or questionable activity is identified and the customer is immediately alerted to improve account security.
Credit underwriting
Machine learning algorithms can be trained on millions of consumer data and financial lending results. What’s more, machine learning can detect the underlying trends that might influence lending in the future. For example, are more young people get into car accidents in certain cities? Are the increasing rates of default among a specific demographic population? This information can be very valuable for financial companies to set their policies. Some early adopters already saw success in this area.
• Micro-targeted model
A European bank replaced older statistical-modeling approaches with machine-learning techniques and reported significant results that they were able to build micro-targeted models that more accurately forecast who will follow credit rules vs. who will cancel service or default on their loans, and how best to intervene. No doubt, this predictive power of machine learning is going to significantly enhance the bank’s future credit underwriting.
• Contract intelligence platform
Contract intelligence platform using machine learning designed to “analyze legal documents and extract important data points and clauses” has been put in used in big banks. Manual review of 12,000 annual commercial credit agreements normally requires approximately 360,000 hours; results from an initial implementation of this machine learning technology showed that the same amount of agreements could be reviewed in seconds.

Pharmaceutical

Discovering a new drug is a long, expensive and often haphazard process for pharmaceutical companies. Thousands of compounds are subject to a progressive series of tests, and only one might turn out to be a viable drug.

Recently, more and more researchers are trying to leveraging AI technology and machine learning at all levels of drug development, from discover new drugs, improve the clinical trial success rate, find new uses of previously tested compounds, create personalized medicine, to find combination medicine.
Discover new drugs
The drug development process, from initial discovery to the marketplace, takes at least ten years. And the success rate for drugs is remarkably low. AI may help to end the pharmaceutical industry’s decade-long struggle with productivity in R&D. According to one national laboratory, uses of AI can potentially cut down drug development time from more than ten years to a single year in some cases.

AI systems trained on various data sources, including preclinical data sets, can help to make significant improvements in new drug development, by finding new biological discovery, discovering new compounds, predicting toxicity and chemical reactions, and eventually improving the drug candidates success rate in clinical trials.
• Find new biological discovery
A new biological discovery provides insight into how the human body or harmful bacteria functions. It’s the first step of new drug discovery.
In this example, AI system first identifies genetic and other markers among sick and well persons by drawing on detailed medical histories, as well as data from scientific publications and chemical databases. It then ranks the genes, proteins or metabolites it finds according to their relevance to a particular disease and determines when specific genes or proteins are associated with certain patient outcomes. Based on the findings, The company will then examine numerous compounds to find ones that may act on a specific biological discovery. Such screening method is “at least 50% cheaper” than traditional methods, according to data scientists in a pharmaceutical company.
• Discover new compounds
Once biological discovery is identified, AI and machine learning can help to generate bispecific-small-molecule compound designs for new drugs.
The machine learning algorithms trained on the vast clinical database can design new molecules that may work to cure the identified biological discovery and is intended to be able to predict better which molecules would be the most effective and safest. The new compounds discovered by AI will have a higher probability to pass the subsequent clinical tests so that pharmaceutical companies can focus their efforts on testing the right drug candidates. Compared to traditional methods, AI and machine learning system can deliver drug candidates roughly 25 percent faster and 25 percent cheaper.  
• Predict toxicity and chemical reactions in preclinical tests
AI system and machine learning can sift through vast amounts of new and existing genetic, metabolic and clinical information to unravel the complex biological networks that underpin diseases, much faster and more accurate than humans. The deep pool of information that AI system injects can help to better predict the likelihood of toxicity in preclinical tests before the company even tries to take a drug to clinical trials. The improvement of screening compounds benefits the entire downstream development timeline and save the company millions of dollars.
AI system can also model chemical reaction on algorithms by learning the “syntax” of reactions. It can predict the correct outcome 80 percent of the time, according to an AI company. While that’s not perfect, it’s nonetheless an incredibly useful tool for cutting down on the amount of time required to research millions of chemical reactions.
• Improve drug candidate success rate in clinical trials
AI systems can better predict in clinical trials what drugs may work for target patients, while simultaneously steering away from drugs that are likely to fail, hence help to improve the drug candidates’ success rate in clinical trials.
Several companies in the U.S. and Europe invite patients from different age and medical situation to form a large database. They collect those patients’ sample of blood, urine, and saliva, then send to a lab where computers using AI and machine learning to scour the samples and genes for molecular fingerprints or biomarkers. This information could later be used to help measure a specific drug’s impact on certain patients, and to identify patients in which such a drug is likely to be most useful. In this case, the big difference between AI-driven drug testing and traditional ones is that companies are not making any hypotheses up front for clinical trials; instead, they are using the patient-derived data to better monitor results.
Find new uses for previously tested compounds
Instead of just trying to discover a new compound from scratch, AI system can also be used to try to find new potential uses for existing compounds.

Numerous compounds went through significant early testing but were abandoned when they were found to be ineffective for the disease they were designed to treat. AI and machine learning can be used to find other diseases for which these drug candidates might instead work. This can dramatically shorten the drug discovery process since all the preliminary safety testing was already done and “learned.” This would allow the new drugs to go directly to clinical trials faster.
Create personalized medicine
The rise of personalized medicine is steering drug developer away from a one-size-fits-all model. Tailoring treatment to individuals’ genes is at the core of so-called personalized medicine.

Machine learning can play a key role in determining what genes and genetic markers are interacting with a possible treatment. In 2017, FDA approved the first treatment only for patients whose cancers have a specific genetic feature. Only a small percentage of patients’ cancers have these specific genetic features, which would make it only effective for this small subset. Machine learning algorithms with a large amount of clinical information at the individual level and always updated can help to identify what medicine may be most effective for certain individuals.
Find combination medicine
Finding combination medicine is even more challenging than looking for single drug treatment. AI can help to find the non-obvious connections that what drugs could work together as combination medicine for better treatment.

The average human researcher reads between 200 and 300 science articles a year while nowadays AI-empowered system can ingest 25 million Medline abstracts, more than 1 million full-text medical journal articles, 4 million patents and is regularly updated. This deep pool of information enables the predictions of combination medicine performance.

For example, the best way to get the body to fight a tumor is some combination of agents to spur the immune system into action. But possible combinations are countless, so AI system trained with historical data is used to tackle the great challenge to find ways to narrow the field and predict the winning combinations that might be most effective for the disease.

Environmental

The environment is a hot topic, as growing demand for resources is leading to land-use changes, climate change, loss of biodiversity, and pollution. The good news is, machine learning and AI technology are uniquely adapted to helping with these sustainability challenges, from finding patterns and interconnections within macro datasets, to providing local, personalized diagnosis and predictions that learn and improve over time.

Businesses are more empowered today than ever before to provide a brighter and greener solution for their business and the world. In turn, by detecting and acting on environmental issues faster and providing more sustainable choices for consumers, they will grow their competitive advantage. It’s a win-win proposition.

Here are a few use cases that illustrate how AI can help resolve the environmental challenges.
Conserve natural resources
By combining satellite imagery, sensors, and machine learning, companies and governments are more capable of conserving natural resources such as water and soil.
• Improve water consumption
The state of California declared a state of emergency in early 2014 due to the ongoing drought and called for a 25 percent reduction in water consumption. Scientists at a geospatial services company provided a solution leveraging machine learning technology to pinpoint water consumption at a land parcel level. That would enable water authorities to target property owners who were using above-average amounts of water for conservation messaging. Meanwhile, machine learning can also help with precision irrigation. In the past, events like extended droughts or flooding required human intervention to assess what happened in the field and adjust the irrigation model based on this new data. By combining satellite imagery and machine learning, a winery in California was able to create an irrigation system that can deliver water in a way that’s situational, hyper-local, automated and self-tuning, helping it cut water use by 25 percent over three years. What is more, this model will adapt and get smarter over time, leading to more water savings.
• Protect soil health
Soil health is impacted by a wide variety of variables such as moisture, salinity, pH value and temperature; and each of these variables is affected by some environmental factors such as ocean water level, acid rain, etc. One company developed a machine learning model that couples weather forecasts with the soil measurements to better predict salinity levels and water absorption rates to provide forecasted soil conditions. In addition to advancing sustainable land-use management, it is also providing agricultural advice to farmers, such as when and how much to fertilize for optimal growth based on current and forecasted soil conditions, to conserve soil health when maximizing production.
Detect and fight pollution
Advanced machine learning is helping organizations to pinpoint the sources of pollution faster and more accurately. This enables more targeted mitigation actions that are better for business and the environment.
• Detect pollution
Air pollution is a major concern in China. Because the previous system was unable to pinpoint sources of pollution at a sufficiently granular level, the government had to take drastic measures to protect air quality, such as factory shut-downs or traffic restrictions. To improve air quality, the Beijing Municipal Government has developed a new solution, which addresses these challenges by using advanced machine learning to identify smaller sections of the city that are at risk, as well, what percentage is from industry versus traffic. Along with trade-off analyses, this enables more targeted mitigation actions, such as only shutting down targeted industries, while minimizing socioeconomic disruptions. In addition to providing more granular predictions, seven-day forecasts are now available 72 hours in advance, giving the city more time to plan and communicate necessary actions and help citizens plan their lives better. Since the solution automatically tunes itself every day from the new data coming in, it becomes more accurate over time as well.
• Fight pollution
Pollution is caused by a variety of human and non-human factors, such as natural disaster, traffic, construction, industrial emission, etc. Which means, pollution in different areas requires different solutions. There is no one-size-fits-all model to fight pollution; it needs to be tailor-made based on the local situation. However, since the combinations of pollution factors can be countless, the solutions that have counter-effect on the pollution factors have countless combinations as well.
Machine learning and AI technology that trained with a vast dataset based on worldwide and local environmental data and past pollution solutions can be used to create the simulation model, to provides the best solution to fight pollution when minimizing the impacts on the ecosystem. This is extremely useful especially when pollution is evolving the machine learning model can also evolve with the new dataset to pinpoint the issues.
Provide sustainable options
More and more business will provide sustainable options for their products and services, to appeal to a growing market of buyers seeking sustainable choices. AI and machine learning can empower business to do more in this area.
• Plan renewable energy production
One of the biggest barriers to widespread use of renewable energy has been forecast accuracy. By combining advanced weather forecasting models with machine learning capabilities, it changes the game completely.

A power company developed a more precise, automated renewable energy forecast based on machine learning for solar and wind power. It enables situational forecasting that is intelligent, automated and self-tuning. On the demand side, for example, they are using a similar coupling of physics and cognitive methods to estimate how much energy will be needed on a given time on a given day, under certain weather conditions, at specific locations. So that business can plan their power production much better.

• Assist with environmental regulation compliance
Machine learning can assist with environmental regulation compliance — an important first step toward greater transparency and greener product choices for consumers. It’s very difficult just to identify all the environmental regulations around the world and understand how they affect business operations. And if you don’t know what’s required, it’s hard to comply with those laws. Machine learning platforms can help humans identify and analyze regulations more efficiently. These capabilities generate information that can help companies gain competitive advantages, and empower consumers to take personal actions that help protect our environment.
Protect nature’s ecosystems
AI and machine learning can be well used to monitor real-time environmental changes, identify and protect species, and predict natural disasters.
Real-time environmental monitoring
It is not always clear how a single stressor, such as salt runoff from roads, affects a natural ecosystem, let alone multiple stressors. In the past, environmental assessments are often manually collected over time, making it very difficult to monitor cause and effect.

To solve this problem, scientists are using machine learning to analyze data from environmental sensors to build and refine computer models of the ecosystem, to better understand the physical and biological processes. Over time, as more data is collected, machine learning will enable a better understanding of how stressors such as salt runoff from roads, invasive species, land-use changes, and climate change impact the ecosystem. This enables decision-makers to conduct real-time environmental monitoring and more targeted remediation, and provide insights to policymakers about the economic impact of their actions such as road treatment practices.

• Species identification and protection
Species are becoming extinct at a rapid pace, which will have devastating impacts on the ecosystem on which humans rely. What is more, what we don’t know is even more worrying – today we only know roughly 1% of all species on Earth. Fortunately, AI technology and machine learning can close this information gap cost-effectively and time-efficiently.
Traditionally, one of the activities in wildlife conservation is monitoring species with camera traps, which take hundreds of shots around the clock; once images are collected, they then need to be analyzed by filling in tables on the type of animals recorded. This rather tedious process can now be automated: we can teach machines to detect and classify flora and fauna from camera trap videos and images. Early work is proving that algorithms can sift through massive amounts of data streaming back from the monitoring systems, and humans and machines can begin to identify the species captured by these remotely deployed cameras and microphones. Automated animal identification performs at the same 96.6% accuracy level of human volunteers, saving approximately 8.2 years of human labeling effort on a 3.2- million-image dataset, according to a research institution.

• Natural disaster predictions
Predictions on major changes in our ecosystem are extremely valuable. To understand and predict where, how, and why environmental changes happen could help us mitigate one of the main threats to our planet. Machine learning can also help to create early warning systems that help us to respond to large-scale natural disasters – tsunamis, hurricanes, tornadoes, flooding, earthquakes, to name just a few.

The National Science Foundation is using machine learning to create a 3-D living model of the entire planet, called EarthCube. The digital representation will combine data sets provided by scientists across a whole slew of disciplines — measurements of the atmosphere and hydrosphere or the geochemistry of the oceans, for example — to mimic conditions on, above, and below the surface. Because of the vast amounts of data the cube will encompass, it will be able to model different conditions and predict how the planet’s systems will respond. And with that information, scientists will be able to suggest ways to avoid catastrophic events or simply plan for those that can’t be avoided, such as flooding or rough weather, before they happen.

Telecom

In the digital era, the telecom industry has shifted from basic phone and internet service to a sector that is going high-tech and constantly evolving into a more mobile and automated environment.  The continuous increase in network size, traffic volume, service complexity, and customer expectations is compelling telecom companies to apply new technologies.

AI and machine learning will enable telecom companies to improve network planning and operations, enhance customer experience, mitigate fraud, and generate leads for sales.

With the industry’s large and spread-out infrastructures, telecom companies also have the opportunity to gather huge amounts of data to feed into AI and machine learning system that fuels the better interfaces with customers.

Here are a few use cases that illustrate how AI can be used to help Telecom industry to grow.
Improve network operation
The demands on operator networks continue to grow with fast adoption of smartphone, VR, AR, and more. To keep pace, network operators must supplement today’s human-centric trouble-shooting and manual remediation methods with machine-based decision making and auto-remediation approaches to maintain and optimize networks at a lower cost structure.
• Predictive maintenance
Predictive maintenance is a major AI initiative in the telecom industry. AI and machine learning can fix problems with telecom hardware (such as cell towers, power lines, etc.) before they happen, by detecting and analyzing signals that usually lead to failure.

A big telecom company in the U.S. started the testing of a drone to expand LTE network coverage in the form of a Flying COW (Cell on Wings) and is exploring ways to incorporate AI and machine learning for the analysis of video data captured by drones for tech support and infrastructure maintenance of cell towers.

Ten years ago, the company would need to send field workers to company sites to check up on hardware periodically – and the company may not know if there was a problem with a specific tower or transformer unless it broke down and needed to be repaired.  Now, AI can detect and analyze the signals and behavior of different “nodes” within its network (e.g., a cell tower) that tip the company off to an impending problem, and they can send a human along to repair it in advance so that customers never notice an outage at all.

In this case, AI and machine learning-based monitoring system can predict faults before they adversely impact network performance, and to guide automated recovery processes.

• Self-optimizing networks (SON)
With AI and machine learning, operators now can set network’s goals and limits. And the network’s machine learning algorithm works within those boundaries to make the network as efficient as possible.

Machine learning can enable the creation of networks that can adjust services based on user need, environmental conditions, and business goals. The machine learning system will learn from past experience as well as real-time network data, to proactively configure networks to meet demand, enable dynamic resource allocation, analyze and optimize network traffic in real time, and therefore improve network usage and reduce costs.

In this case, AI is used to create a cognitive management architecture with little to no human interaction, to continuously optimize network performance.
Enhance marketing and provide better customer experience
The big data and AI platform offer the company and its users the ability to take the data they currently collect and use it towards activities such as personalized marketing campaigns, laser-targeted advertising, and deep customer engagement.
Mitigate fraud
Fraud is a major issue within the telecoms industry, directly impacting the bottom line. In the past, most network operators utilized fraud mitigation tools which use rules-based logic to identify fraudulent behavior.

Now, more and more operators will explore machine learning to develop models that can automatically identify activities as fraudulent or non-fraudulent, based on connections to large datasets such as customer profile, real-time activities, social media feeds, and behavior pattern. Machine learning may also offer the potential to spot early trending anomalies associated with criminal behavior.

Renewable Energy

The renewable energy is a growing economic force and an effective strategy towards better environmental sustainability.

Despite the increasing use and awareness of renewable energy, there is still resistance to wider implementation. Companies are exploring how artificial intelligence could assist in improving accessibility and efficiency of renewable energy, and improve overall energy consumption as a whole.

Here are a few use cases about how the energy sector can benefit from AI.
Improve renewable energy production forecasting
Weather can often be unpredictable, destabilizing the power supply generated from weather-dependent energy sources such as solar and wind. This puts pressure on the renewable energy sector to efficiently balance supply and demand. In fact, a consistent challenge with wind and solar power is their unreliability, which may cause suppliers to rely on traditional energy sources to meet consumer demands.

Therefore, today, companies are incorporating artificial intelligence to improve the accuracy of renewable energy forecasting, to respond to weather fluctuations that may negatively affect operations proactively and to plan supply accordingly. The AI-driven energy forecasting system mines a combination of data from local satellite images, weather stations reports, and wind farms data in the surrounding area.  The machine learning algorithms trained on heavy industry data can generate weather forecast models. Based on the combined data sets and the weather forecast models, the algorithms can then conduct weather forecast for a particular area at any given time. Such forecasts can range from hours to days in advance with data updates several times a day.

With the more accurate and more frequent updates on weather forecast by machine learning algorithms, greater precautions could be taken to harness and preserve the renewable energy generation.
Manage energy production from multiple sources
Renewable energy and conventional power will probably co-exist for a very long time. It is critical to manage energy production from these multiple sources to meet demand requirement when encouraging the use of renewable energy when we can. AI and machine learning can be used to automate the process and balance the demand and supply.
• Manage demand
On the demand side, smart meters for homes and businesses and sensors along transmission lines will be able to collect data in real time and send it to the machine learning algorithms, allowing operators to manage energy production and avoid disruptions actively.

For example, these sensors would communicate with the grid and modify electricity use during off-peak times, thereby relaxing the workload of the energy production during those hours, and subsequently lowering prices for consumers and reduce greenhouse gas emission as well.

• Manage supply

On the supply side, AI will allow the transition to an energy portfolio with increased renewable resource production.
For example, when renewable energy production is operating above a certain threshold, either due to increases in wind strength or sunshine, the grid would reduce its production from fossil fuels, thus limiting harmful greenhouse gas emissions. The opposite would be true during times of below-peak renewable power generation, thus allowing all sources of energy including fossil fuels to be produced.

Additionally, producers will be able to manage the output of energy generated from multiple sources to match social and temporal variations in demand in real-time.
Modernize energy delivery network
In modern times, energy demand continues to rise; AI will be the brain of the future energy delivery network, which monitors and controls every consumer and node, ensuring a two-way flow of electricity and information.

In the U.S., millions of smart meters that monitor energy usage per device and alert utilities of local blackouts have been installed. The AI-power system will continuously collect and synthesize overwhelming amounts of data from those smart sensors nationwide to make timely decisions on how to best allocate energy resources. Additionally, the deep learning algorithms will continue to learn on their own from spotting patterns and anomalies in large data sets, to optimize the delivery model. It is estimated that while total U.S. energy demand is expected to increase 25 percent by 2050, the AI program will limit the rise in peak electricity load on the grid to only 1 percent, thus minimize the risks of the massive blackout.
Lower the costs of renewable energy
AI and machine learning are being used in a combination ways to lower the cost of renewable energy from development to sales so that it can compete with fossil fuel energy.

For example, A British startup which sells solar panels and batteries is using machine learning to learn its customer's usage patterns and to manage the batteries and power sources most optimally. Based on which, they improve the product development: if a customer’s battery starts to get low, the system automatically adjusts the energy consumption such as brightness of the lights or the rate of cell phone charging, to make the energy last as long as possible and allow time to re-produce energy. Such solar product design can lower the renewable energy costs, thus making it more attractive to customers.

In another example, renewable companies are using machine learning algorithms to lower sales costs. Selling solar panels to customers is an expensive problem for companies, and the high sales costs will be eventually transferred to customers, which make the energy product less attractive. AI and machine learning can analyze vast data sets such as industry data, current and potential customer profiles, weather patterns, energy consumption data, etc., and help to predict which prospects may buy solar panels, what may be the most effective personalized marketing campaign, and what is the optimal offer for those prospects.
Optimize energy consumption
AI and machine learning can be leveraged to optimize energy consumption. This can be particularly helpful for large commercial buildings to improve energy efficiency and save money.

The process begins with the installation of IoT hardware. Smart sensors are directly attached to the client’s electrical circuits to track energy consumption data. Since every appliance has a unique electrical footprint, the machine learning algorithms can identify each unique energy source while still providing a comprehensive analysis of the data captured by all the smart sensors. The algorithms read and analyze the data on a real-time basis, and inform the building manager how the building consumes energy, what is breaking down or not working properly thus causing energy waste, where and when energy is consumed higher than normal, what is costing the most energy consumption. What is more, the machine learning algorithms are also “trained” by different operating scenarios and parameters, so it “understands” how buildings of different sizes and designs function, and can help to identify opportunities for optimization.

In one case study, a hotel in San Francisco identified energy inefficiencies in the hotel’s commercial kitchen, within a three month period, AI system reportedly identified inefficiencies that were costing the hotel more than $13,000 in preventable energy costs.

More to come soon!

There are countless industries that are currently adopting and exploring AI; we are working hard on defining more educational material about their adoption use cases as you are reading this!
Get in Touch
© 2017 R2.ai, inc. All Rights Reserved
x
Work with us
* All fields are required
Thank you! Your submission has been received! If appropriate, a member of R2.ai's business team will soon be in touch to both introduce you to our company as well as learn more about your unique needs. We look forward building a relationship with you.
Oops! Something went wrong while submitting the form. Please make sure that all fields are filled out correctly.