In April, hackers successfully breached the networks of a Canadian gas pipeline. Once in, they were able to increase valve pressure, disable alarms, and make emergency shutdowns. An attack at this level is unprecedented, but the reality is the energy sector is increasingly exposed to malicious players.

One report already shows that the energy industry is a top target for cyberattacks. Energy is essential to communities, and part of the reason that the sector is so attractive to criminals is that they hold substantial leverage if they compromise networks. In 2015, for example, a hack on a power grid in Ukraine saw a quarter of a million inhabitants lose electricity.

• In 2022, 10.7% of observed cyberattacks targeted the energy industry.
• More than 75% of energy companies are vulnerable to ransomware attacks because their account credentials are readily available online.
• Cyber attacks cost the energy sector $4.72 million per breach on average in 2022.

The risk posed to the energy sector is exacerbated by its infrastructure. Much of the operational and informational technologies responsible for running energy systems haven’t been engineered for today’s digital environment — and this disconnect creates gaps in energy systems that hackers can take advantage of. Moving forward, energy organizations need to update their defensive toolkit to more accurately identify vulnerabilities, detect foul play, and strengthen authentication steps.

It starts with artificial intelligence.

Strengthening the overall energy cybersecurity chain

The energy sector’s defense against hacks is only as impactful as its weakest contributor. All players need to implement smart solutions that construct a collective front and ensure inefficiencies are addressed quickly. With a stronger security net over the entire energy sector, hackers won’t be able to carry out attacks on a large scale and therefore lose a lot of their leverage.

Using AI, energy companies can detect and monitor threats in their operating technologies. Insights can also be shared across companies, helping educate organizations about emerging attacks and how to thwart them. AI can also process the huge swaths of data that energy companies have and generate valuable outcomes for cybersecurity. For example, data around system usage, historical performance, and behavioral trends could suggest when an attack would happen and at what localized point of the network. Machine learning algorithms can equally be trained to recognize patterns and anomalies in data, and in response, can alert an operator or even automatically shut down designated system components.

As AI becomes more affordable and versatile, energy companies of all sizes can apply and scale it within their business. And, with more players integrating AI, takeaways around cybersecurity will be more informed and more effective as a protective measure.

Automated AI for the highest level of protection

Automation is at the core of good cybersecurity. Energy companies need to know that they have high-quality defenses in place on a continuous basis. Rather than spend large amounts of time on manual reviews, AI can keep atop of cybersecurity maintenance by providing real-time, autonomous protection against the likes of zero-day threats (which exploit bugs or access in software).

This automated AI doesn’t affect efficiency in energy companies’ workflows either — there is no downtime in processes, and issues due to human error are removed. Open AI is a prime example. Language models like the ones at Southern California Edison can be used for various applications in the field of cybersecurity, including threat intelligence, log analysis, and real-time incident response, all without interrupting daily technical operations.

What’s more, automation can be the bridge between legacy hardware, remote assets, and new technology in the energy sector. Instead of having these resources operate in silos — and create more entry points for hackers — AI offers a unified shield, and can be retrofitted to ensure that all equipment has the most up-to-date cybersecurity. Especially among legacy hardware, which is more likely to break and can be expensive to replace, automated AI solutions bring the equipment’s defenses up to par with modern hardware.

AI considerations for comprehensive security coverage

AI is evolving, and so too is the energy sector’s application of it. Companies can’t assume that they can introduce AI and take a step back from cybersecurity — they need to stay atop of challenges in order to be better prepared to respond to them. The result is more streamlined security in the most pressing moments.

For instance, energy organizations need to track and carefully select the data that they use to fuel AI. The quality of data is far more important than quantity of data, and companies should take care not to fuel AI with information that could produce misleading or inaccurate conclusions and negate cybersecurity actions. That’s why energy firms need dedicated AI teams that review how AI is driven and where data can be refined.

Energy consumption is also a factor in AI application. AI consumes a lot of electricity, and in the energy sphere where companies understand the importance of efficiency, players have to balance essential and advanced cybersecurity with a sustainable financial and environmental impact.

Meanwhile, compliance has to be a priority for AI adopters in the energy sector. Regulation around AI is still developing, and governments are set to implement greater structure around it in the future. Organizations subsequently need to stay aware of their legal and ethical obligations and take an agile approach to AI that enables them to adjust how they use it, without losing its benefits.

Cybersecurity lapses in the energy sector can have severe repercussions for the companies themselves and the general public. It is urgent that AI be used to safeguard energy infrastructure, as well as to shape general practices that boost the health of energy firms. Organizations that embrace AI now not only heed the warning from the Canada pipeline hack, they establish more robust cybersecurity for the long term.

Published On: Utility Drive

By now, artificial intelligence is a cog in the well-oiled manufacturing machine. In fact, the AI in manufacturing market is predicted to reach a staggering $16.3 billion by 2027. How has it reached such heights? AI is a key element of the unfolding Fourth Industrial Revolution, being able to track core KPIs, produce accurate forecasting reports, anticipate tech disruptions, detect inefficiencies in real-time, and much, much more. That’s why the likes of BMW, Nissan, Canon, and Boeing have incorporated AI into their manufacturing processes.

But AI isn’t reserved for big players. Smaller companies are also leveraging AI to transform their operations, be more cost-effective, and optimize their sustainability efforts. Especially as some of the biggest concerns facing manufacturing at the moment include economic uncertainty, supply chain issues, and pending legislation around environmental practices, AI is a necessary solution for companies to stay resilient and agile.

Reports show that companies using AI have seen notable cost savings and revenue growth, but AI makes more than financial sense — it shapes companies that are more conscious of people and the planet. Here’s why AI adoption goes beyond four walls, and why every manufacturer should be integrating it.

More efficient product development across the board

AI functions across the entire lifecycle of a product. To begin, AI systems use machine learning to identify buying patterns and translate them into insights for manufacturers. For example, Danone Group uses AI for its demand forecasting and has seen a subsequent 30% reduction in lost sales, a 50% reduction in demand planners’ workload, and a 20% reduction in forecasting errors. Such data enables manufacturers to scale manufacturing accordingly and build products in more efficient ways.

AI software can also create multiple optimized designs for a product based on parameters like materials, size, manufacturing methods, and cost. With these designs, manufacturers can move forward with product development that is less time and resource-consuming. The effectiveness of AI-powered designs is why carmaker Nissan is currently developing its own AI to design cars without any human input.

Elsewhere, AI robotics and cobots (collaborative robots) facilitate product assembly. These machines work in close proximity to humans and can pick up, place, and sort through objects in the manufacturing process. AI improves their orientation and precision, meaning that more complex products can be built, and at a faster speed.

Perhaps one of the most impactful areas of AI in product development, however, is quality assurance. AI can ensure that items are fit for sale and meet companies’ high standards of delivery. Apple, Nintendo, Nokia, and Sony use Google Cloud Visual Inspection AI to find defects in their manufacturing, as the technology can highlight wrong, misplayed, missing, rotated, or deformed components at various stages of the assembly process. Even better, using the AI tool requires no previous technical expertise, so smaller, growing companies can embrace it too.

Other AI software called Robotic Process Automation can carry out repetitive, high-volume tasks like updating records, addressing queries, and performing calculations.

Smarter, safer (virtual) factories

Manufacturers have an inherent responsibility to protect their workers and warehouses — and they can do so using cutting-edge AI.

For one, AI programs can analyze data at a quicker and more accurate rate than humans, spotting dangerous situations in real time and alerting the right people to take action. Algorithms can comb through camera footage from manufacturing sites and flag high-risk situations. Likewise, AI can detect non-compliance like not wearing a helmet or harness in factories — which are some of the leading causes of accidents in manufacturing.

AI wearable sensors worn on clothing also contribute to safer work environments. These devices can monitor activities and stream data, if anomalies appear in the data, managers can be alerted to potential accidents; for instance, if large groups of workers are crowding around falling hazards.

The same heightened safety awareness applies to digital twin models, which are AI-driven, virtual representations of manufacturing floors. The models give manufacturers greater scope to experiment with scenarios in warehouses and make informed decisions about what technology and protocols to implement. Digital twins can essentially simulate operations and generate evidence to support safety cases — for example, by proving that robotics can weld materials and safely pass the completed component to a staff member for further assembly. It’s no surprise then, that already, one in five manufacturers is experimenting with actively developing a metaverse platform (a type of digital twin) for products and services.

Fueling green corporate responsibility

Corporate responsibility is a form of self-regulation where manufacturers undertake measures that benefit both people and the planet — and it’s becoming standard practice in the industry that produces one-fifth of all carbon emissions.

AI plays a big role in helping manufacturers be more sustainable. Predictions suggest that the tech could contribute to 79% of the UN’s Sustainable Development Goals, as well as accelerate greener behaviors across the board in manufacturing. For one, AI can help drive the reusing of heat in factory buildings and regulate lighting according to the number of people in the space. This management of facilities dramatically minimizes energy losses.

AI-based applications can also predict energy consumption, showing manufacturers where machines are inefficient and thus pushing them to invest in more eco-friendly hardware. AI solutions for one automotive company found that 40% of energy consumption for one machine occurred when it was not producing anything; the insight prompted the company to power down the machine more often and reap impressive energy and financial savings.

As stricter legislation is introduced to carry out sustainable practices in manufacturing, AI can guide businesses around their most energy-consuming areas and equipment, and prompt them to make positive eco changes before it becomes a mandatory (and expensive) switch.

Artificial intelligence has become synonymous with manufacturing — and not just in building physical products. It is facilitating efficiency, safety, and sustainability as a whole in the sector. And yet, we’re still in the early days of realizing what AI can do for the industry. Manufacturing companies that haven’t embraced the potential of AI need to do so now to reap the AI benefits today, as well as the AI innovation of tomorrow.

Published On: AI Journal

Originally Published in TechCrunch on April 12, 2021

As AI technologies become more advanced, previously cutting-edge — but generic — AI models are becoming commonplace, such as Google Cloud’s Vision AI or Amazon Rekognition. While effective in some use cases, these solutions do not suit industry-specific needs right out of the box. Organizations that seek the most accurate results from their AI projects will simply have to turn to industry-specific models.

There are a few ways that companies can generate industry-specific results. One would be to adopt a hybrid approach — taking an open-source generic AI model and training it further to align with the business's specific needs. Companies could also look to third-party vendors, such as IBM or C3, and access a complete solution, off-the-shelf. Or — if they really needed to — data science teams could build their own models in-house, from scratch.

Let’s dive into each of these approaches and how businesses can decide which one works for their distinct circumstances.

Generic models alone often don’t cut it

Generic AI models like Google Cloud’s Vision AI or Amazon Rekognition or open-source ones from Tensorflow or Ski-kit-Learn often fail to produce sufficient results when it comes to niche use cases in industries like finance or energy sector. Many businesses have unique needs, and models which don’t have the contextual data of a certain industry will not be able to provide relevant results.

Building on top of open-source models

At ThirdEye Data, we recently worked with a utility company to tag and detect defects in electric poles by using AI to analyze thousands of images. We started off using Google Vision API and found that it was unable to produce our desired results — with the precision and recall values of the AI models completely unusable. The models were unable to read the characters within the tags on the electric poles 90% of the time, as it didn’t identify the non-standard font and varying background colors used in the tags.

So, we took base computer vision models from TensorFlow and optimized them to the utility company’s precise needs. After two months of developing AI models to detect and decipher tags on the electric poles, and another two months of training these models, the results are displaying accuracy levels of over 90%. These will continue to improve over time with retraining iterations.

Any team looking to expand its AI capabilities should first apply its data and use cases to a generic model and assess the results. Open source algorithms that companies can start off with can be found on AI and ML frameworks like TensorFlow, Ski-kit Learn, or Microsoft Cognitive Toolkit. At ThirdEye Data, we used Convolutional Neural Network (CNN) algorithms on TensorFlow.

Then, if the results are insufficient, the team can extend the algorithm by training it further on their own industry-specific data.

Let’s take an example. A small to medium-sized tax services company based in Los Angeles is leveraging AI models to power its customer service chatbot. A more generic chatbot feature would be limited in its ability to answer specific domain-related questions. However, once they train it on up-to-date, industry-specific data, such as federal and state-level information on IRS processes and forms, the model will be able to accurately respond to specific customer questions around filing taxes in 2021 and all the caveats that it brings.

Deploying off-the-shelf industry-specific AI solutions

As the need for more tailored AI solutions grows, so do the market opportunities to leverage models that have been trained on industry-specific data. Dozens of these offerings now exist, with many of them coming from Big Tech and AI companies, and we should expect to see that number explode over the next five or so years.

For example, IBM has a complete list of industry-specific AI models, including Watson solutions for customer service, supply chain, financial operations, and security — to name but a few. IBM also recently collaborated with Atos to build industry-specific automation solutions that use AI and hybrid cloud technologies.

C3.ai, the enterprise AI company, has recently released a new suite of industry-specific AI solutions, including ones for energy management, anti-money laundering, and production schedule optimization. The organization has an impressive 4.8 million AI models in use which, according to the company website, provide “powerful predictions across the most complex organizations.”

H2O’s solutions range from financial services and insurance to manufacturing and retail. In the retail sphere, the company’s AI models have been able to help leading brands like Macy’s, Walgreens, and eBay forecast product demand, boost CX through personalized experiences, and drive advanced inventory planning.

While developers don’t get full access to the model itself, the investment in these types of solutions gives them access to many other valuable services that they can use while implementing the AI models. For example, they might need assistance with setting up secure data pipelines — a necessary feature given how many data sources a team may be handling at a given time. These services might also include help with preparing the data, data privacy and security assurance, or version management of the AI models.

So, how do companies know which is the best approach for them?

Choosing the right approach

When selecting the best way forward to deploy industry-specific AI solutions, it’s important to keep a few things in mind.

Companies deciding whether to build on top of an open-source model should note that it’s easier to expand on the existing capabilities of AI models than building them from scratch. With the original model available as open-source, the end product belongs to the data science team to continuously tweak and develop.

However, they should remember that it’s necessary to build on and train the model in an environment that lets them do so. Azure ML, for example, is a development environment that lets you import an open-source algorithm and build a model on top of it.

Still, taking the OS approach can come with some setbacks. If models are running across different clouds, that can be time and resource-consuming to manage. And inevitably, there will be no guarantee of the model’s performance or service level agreement, as it’s not a paid service — the data science team managing and training the model must stay on top of maintenance.

If a business is seeking guaranteed performance and a full stack of accompanying services, they might consider leveraging an off-the-shelf tool.

Investing in these pre-trained AI models is a good idea for companies looking to deploy AI models for an extended period of time that require scalability and reliability. These are often enterprises that have their AI needs largely figured out and require the guarantee of service level agreements.

Though, with the AI models ultimately residing in the vendor’s environment, developers working with the model won’t be able to take those AI projects elsewhere. That’s why opting for an out-of-the-box solution should be a long-term commitment, for companies who are prepared to invest in an advanced solution at the expense of ownership over the model.

And for a select few organizations, building from scratch might actually be the best option — though, given the amount of time and resources necessary to build an industry-specific solution from zero, it’s only advisable to do this when the other options don’t suffice.

For example, extremely niche industries doing truly novel work such as drug manufacturing or medical research will have to build and train their own algorithms from the ground up. For example, take researchers using AI to identify patterns in genetic code for a certain disease. This kind of use case likely wouldn’t have pre-built models that can provide a base for further training.

Whatever path a company takes will depend highly on the nature of the AI project. Companies should carefully consider their technology and maintenance needs, how much budget and time they are willing to spend, and from there decide on which approach to adopt. One thing is for certain, though: Industry-specific AI models are only going to boom in popularity over the next few years, and businesses from across sectors realize their power in delivering accurate and powerful insights.

The humble search engine has come a long way since its conception in the early 1990s. Yet only recently has it taken serious strides towards giving much more nuanced and relevant responses, with the help of artificial intelligence (AI) technologies machine learning (ML) and natural language processing (NLP).

Incorporating the capabilities of AI into search engines creates semantic search. No longer are searches based on keywords and their dictionary definition: Semantic search means understanding the intent behind the query and representing the “knowledge in a way suitable for meaningful retrieval,” according to Towards Data Science.

Semantic Search uses ML to learn from the results of past queries to further hone and improve accuracy and relevance. NLP comes in to let the searcher phrase their queries in a conversational way, as they would if they were speaking to a human. This allows them to spend less time thinking of the best keywords related to a certain search.

Unlike its keyword-based predecessor, semantic search can process data from a range of sources, including email, social media, documents, PDF., images, video, and audio. This expands the possibilities for the searcher hugely by allowing them to use all of the materials at their fingertips to find what they’re looking for.

Let’s dive into the potential semantic search holds for businesses across the board.

Semantic Search is a worthwhile investment for organizations

There isn’t an organization out there that doesn’t stand to benefit from receiving faster, more accurate, and higher quality results. By excluding irrelevant information and delivering only the most accurate responses, semantic search facilitates less searching, and more discovery.

For example, let’s say an underwater welder is looking for equipment to complete a job on a 1,500ft deep oil rig in the Gulf of Mexico. A keyword search would not understand the multifaceted nature of this job, and might only provide results related to welding. A semantic search would understand the context behind such a project, and understand that the searcher is a diver first, and welder second. It would offer relevant results such as a hyperbaric chamber and take into account factors like working conditions and water currents.

Semantic search can provide other business benefits by combining NLP with a user-friendly interface, making it easy for anyone to interact with and immediately find the results they are looking for. Having quick access to accurate results helps inform decision-making and powers productivity in businesses of all sizes. Semantic search can also unify unstructured data from diverse sources to draw insights that help drive business growth and development.

Where have we seen Semantic Search so far?

Unsurprisingly, Google has been leveraging semantic search to improve its user experience since 2013, when it launched the Hummingbird update. This update introduced “conversational search” into its capabilities, meaning the context of the whole query is taken into account, rather than the individual words.

In 2015, Google launched RankBrain, a machine learning system that also seeks to understand user intent behind searches. RankBrain represented a step up as its algorithm learns by analyzing the best-performing search results, identifying the most accurate response to a search even if it doesn’t contain the exact words used in the query.

Online retailer Zappos recently implemented semantic search into its website to allow its site users to find exactly what they were after, quickly and easily. The company’s chief data scientist explained that the technology not only understand the context of the search term, but also tailors the result to each customer’s past search data. Zappos is hence able to individually serve each of their customers with the results that are most relevant to them.

As the value of semantic search becomes more visible, organizations all over the world will increasingly adopt platforms like that of Microsoft Azure and AWS Comprehend that enable them to leverage the technology without investing in building it themselves.

What should be considered before adopting Semantic Search?

While semantic search will undoubtedly provide an ROI for businesses that adopt and use it with a clear purpose in mind, there are still some things that need to be considered. In order for smooth implementation, the enterprise needs to have a clear sense of what outcomes it is looking for, and then make sure to feed the search engine the appropriate data in the right quantity.

Efficient use of semantic search also means breaking down any data silos that occur within an organization to make sure that no crucial information is missed. By ensuring that employees are open with the necessary data and that there is proper technical planning, organizations will be able to train their semantic search engine to produce the most informed and accurate results.

It might seem like semantic search is a well-developed technology, but in reality there is still a long way to go. But that’s not to say that we should wait until its full capabilities have been realized to start making use of its many benefits: With semantic search, businesses of all types can facilitate faster and more-informed decision making, improve user experience, and ultimately drive their bottom line forward.

ThirdEye Data: The Answer to All Data Challenges

Big data is no fad. The world is witnessing a meteoric rise of data today, which is only doubling in volume by the year. As data evolves, every business organization seeks to explore the deluge of information and glean meaningful insights to drive better decision-making and enhance productivity. Despite such potential, enterprises, especially SMBs, fall behind in implementing data-driven processes. The culprit, however, is not their lack of innovation, but the complexity in unraveling the intricate correlations between seemingly unrelated data. Attending to alleviate these setbacks, an AI-driven data-focused company ThirdEye Data lays its cornerstone.

Founded in 2010, a one-stop shop for data sciences, analytics and engineering services, and products, “ThirdEye is the answer to all your data related queries,” remarks Dj Das, Founder, and CEO of the company. Leveraging modern technologies such as AI and ML, ThirdEye offers actionable insights, realworld experiences, and strategic recommendations to help enterprises mitigate their business challenges.

Further highlighting the robustness of ThirdEye in today’s business milieu, Aparajeeta Das, co-Founder and Chief Delivery Officer of the company, notes, “Even to this day, many firms struggle to deploy digital data gathering methodologies, let alone implement sophisticated data processing algorithms.” One of ThirdEye’s primary objectives is equipping such companies with the right solutions to make them more technologically mature in the era of digitalization.

ThirdEye aspires to become the proponent of digitalization for all enterprises, big and small, all over the world. “Smaller firms are usually underserved when it comes to implementing data solutions,” says Dj. The reasons are manifold: the high cost of deploying a big data solution; lack of trained data scientists to utilize the sophisticated analytics toolsets; or inability of firms to determine the starting point of their digital data journey. “Laser-focused on countering these problems, we combine different big data technologies, like Apache’s Hadoop and Spark with AI and ML technologies like TensorFlow, SparkML, and Azure ML to offer the best possible solution at the best possible price with the least TCO,” says Dj.

Committed to this objective, ThirdEye recently launched a new product named ClouDhiti, which offers “Business Analytics as a Service,” targeted at the SMBs. This offering is both cost-effective and less cumbersome for an SMB. “Currently, we are extending this model to support clients from restaurants, education, and healthcare sectors. We are also utilizing ClouDhiti to help various NGOs on their social missions,” says Aparajeeta, emphasizing the diverse possibilities of their approach.

Alongside, ThirdEye also offers three distinctive products: Eyera, Safera, and Syra. From network fraud detection and predictions in the real-estate sector to predictive maintenance and fleet analytics for the transportation industry, Eyera — a comprehensive IoT solution — can be customized for a variety of business use cases. Safera, on the other hand, is a predictive analytics platform for analyzing criminal activities and predicting the probability of crime in specific locations given the base information about weather and traffic. Lastly, Syra is a customizable AI-driven chatbot solution that can answer questions specific to an enterprise’s domain. These chatbots deliver higher engagement and hence higher conversion ratios.

“Today, data has become the fulcrum on which the axle of a business turns; it is no longer a local phenomenon, but a global trend,” says Dj. Jumping on this bandwagon, ThirdEye — without physical sales offices outside the U.S. — has clients from seven countries worldwide. “The world has become flat when it comes to data,” comments Dj. In the coming years, the duo intends to leverage their decades of expertise in handling the data deluge to make ThirdEye the numero uno in the data landscape. Expressing their passion for redefining the data solutions landscape, Dj concludes with a soliloquy from his favorite Bollywood movie: “What is the point of living, if you do not have an impossible dream to fulfill.”

READ THE ARTICLE ON SILICONINDIA.COM

We are living in an age where convenience, speed, and efficiency are king. So, unsurprisingly, chatbots have taken reign as the preferred method of customer service for many — not least fast-paced professionals that don’t have the time to sit through automated messages or wait on hold for a human operator.

In fact, 69 percent of consumers prefer chatbots for quick communication with brands, and Gartner predicted that by 2020, 85 percent of engagement with businesses will be conducted without the consumer ever interacting with another human being.

There’s few places that AI-powered chatbots have more potential than in the financial sector. Many of the big banks have already introduced chatbots into their customer service infrastructure, including Bank of America, Capital One, MasterCard and American Express.

Yet it’s important to remember that, in finance, chatbots enhanced with AI have much more to bring to the table than just dealing with basic transaction and information-based requests.

AI chatbots can seriously help financial service providers of all sizes in areas like data security, fraud detection, and assistance with wealth management. Here’s how.

Data is safer with a chatbot

While many people may question the integrity of chatbots’ data security, when built and used correctly, they can actually provide more in the way of privacy and protection of the user’s details than a human operator.

Most of the customer support requests that financial institutions receive are related to users’ banking information, their balance, and other private details. This is delicate information that’s crucial to keep secure, so with insider theft of personally identifiable information (PII) proving to be more of a threat than outside hacks, it makes sense to hand the responsibility of keeping it safe with a trustworthy AI-based chatbot.

Chatbots can take over functions involving the exchange of PII, totally removing humans from the loop, thus increasing overall security. The chatbot will dutifully follow all of the data privacy and protection policies that the security engineer enforces and keep to the confines of how it’s instructed to access the data required for the conversation.

Thus, as the data flows between the user, the chatbot and the backend systems — essentially “data-in-motion” — it’s guaranteed to be secure as there are fewer human touch points and a reduced probability of process breakdowns.

By using a chatbot for these conversations rather than a human, banks and other financial service providers can increase security without making the customer do more — something 75 percent of businesses have stated as being a priority for them.

And while most human operators won’t have malicious intent to steal PII, there may still be instances of negligence. Humans will never be 100 percent reliable, and in many cases they will prove to be the weakest link in this type of secured environment. Deploying an AI chatbot is a great way to reduce the risk of data security breaches and give customers peace of mind.

Chatbots track and report on fraud, in real time

Fraudulent activity is undoubtedly one of the biggest problems financial service providers have to face day in, day out. And with 61 percent of fraud losses for major banks stemming from identity fraud, financial services need to find a way to better ensure that their users are who they say they are.

Luckily, chatbots can detect such activity much better than humans can. For example, imagine a user of a chatbot is trying to impersonate someone else, and starts the conversation by providing the correct authentication information for the person who’s identity they want to steal, giving them access to the information that they were seeking.

However, in later conversations the user stumbles on basic questions, such as information on the weather in the correct user’s local area. Further, the user’s responses to other questions don’t match the “writing” profile on the correct user that has been created by the backend AI systems over time.

Let’s also say that the real customer is supposed to be based out of San Jose, CA, yet the incoming IP address is in another country. An AI-based chatbot can detect all of these inconsistencies in real time, alert the financial service provider, and give them the option to verify suspicious transactions and advise on next steps.

In addition to this, AI chatbots with voice and facial recognition capabilities will add another layer to preventing identity theft in financial institutions. Chatbots can analyze the voiceprint of the user in real time, compare it with the previously stored voiceprints and detect if the current user is the real user or not. Additionally, these chatbots can detect tension, nervousness or even malicious intent by analyzing the voiceprints in real time. When detected, they can then alert the financial service provider so they can act accordingly to prevent the PII theft attempt.

Fraud prevention in banking and finance is critical to maintaining customer trust in security practices. Integrating AI-based chatbots into their security infrastructure is a great way to reduce the probability of such incidents.

Chatbots can provide personalized, data-powered advice

Chatbots also have huge potential in helping people manage their money. In terms of investment support, the AI-powered backend system of a chatbot has capabilities to assess a user’s portfolio based on their profile, calculate the risk-taking capacity of the user, and manage their wealth in a much more organized way. The system can then recommend portfolio updates, change the risk profile based on a conversation, largely replacing the need for a human financial adviser.

For day-to-day banking, the AI can understand the customer’s spending habits by tracking their card transactions and give budget and planning tips based on this data.

Now enhanced with AI, Bank of America’s virtual assistant Erica is able to help customers tackle complex tasks and provide personalized guidance to help them stay on top of their finances. Erica gives its 3.6 million users insights such as monthly spending snapshots, flags upcoming recurring charges, and helps to track month-to-month changes to FICO® scores.

Aditya Bhasin, head of Consumer, Small Business and Wealth Management Technology for Bank of America, explained that they have integrated more than 200,000 different ways for the clients to ask financial questions and build Erica’s conversational knowledge.

Clearly, AI-powered chatbots have a lot more to offer the banking and finance sector than just customer service support. And while customer service chatbots have their benefits (not least reducing the need for humans to do simple, automatable tasks) financial service providers — from small fintech startups to big banks — should be leveraging the power of AI to improve their customers’ data security, prevent fraudulent activity, and help their customers make the best decisions they can with their money.

DJ Das is the Founder and CEO of ThirdEye Data, a Data Science, engineering and analytics firm. DJ has successfully delivered new products and services to over 20 Fortune 500 companies. A serial entrepreneur, DJ is also an angel investor in various data-centric startups in Silicon Valley.

Source: Bots For Banking: How Smart Chatbots Can Secure Data And Combat Fraud

It’s no secret that when chatbots are bad, they’re really bad. There’s nothing worse than trying to convey your needs and problems in a chat box, only to be met with irrelevant automated responses.

Even entities such as CNN and the Wall Street Journal have struggled to build bots that could understand simple unsubscribe requests from their users. CNN’s bot only understood the request when the word “unsubscribe” was sent on its own, while WSJ’s bot just ignored the request altogether, and continued to bombard its users.

Customer interactions with rudimentary chatbots can be slow, intrusive, and misleading, and are enough to drive visitors away for good. In fact, studies show that 73 percent of users would not use a chatbot again if they perceived the experience to be negative. A supposedly helpful tool could end up driving revenue away from your business — yet many companies still choose to deploy these clunky, generic tools.

However, with the advent of artificial intelligence (AI) and natural language understanding (NLU) technologies, custom chatbots are able to interact with site visitors in a much more informative and fluid way, providing contextual information and guiding users to their pre-set goals.

Let’s take a look at why companies need to scrap many of the current chatbots and invest in ones that are enhanced with these new technologies.

Where is the traditional bot going wrong?

We’re all familiar with the feeling of getting nowhere with automated responses, only to end up needing to speak to a human customer service representative after wasting 20 minutes of our time with a chatbot. This usually happens with rule-based chatbots that follow strict conversation paths, leading to dead ends where the chatbots doesn’t have any appropriate responses for the questions asked.

Traditional chatbots are often made in a cookie-cutter fashion for any website. For example, a chatbot that’s been deployed on a retail site may be the same one being used on a news site, and thus answer questions in the same fashion without any domain knowledge of either website.

In fact, if the information the user is seeking to find out is simple enough to be given out by a rule-based chatbot, there’s probably a more efficient way to convey that information. And there’s no need to create chatbots to solve problems that don’t require a back-and-forth conversation, like finding out opening hours or buying tickets, for example.

Not to mention, chatbots without NLU can only recognize keywords and act on them, and often give customers the option to choose from a long list of buttons to move the interaction along. All of this significantly decreases the amount of time any user wants to spend interacting with a chatbot — and thus the amount of time they spend engaging with your business.

What difference do AI and NLU make?

The contrast between traditional chatbots and those enhanced with AI and NLU is stark and can be the difference between a customer having an informative, free-flowing conversation that swiftly solves their problem, and them being put off by irrelevant questions and unnatural dialogue.

According to Gartner, AI “appears to emulate human performance typically by learning, coming to its own conclusions [and by] appearing to understand complex content.” This makes AI-enhanced chatbots much more conducive to natural dialogue. Not only this, but AI chatbots improve continuously because of the amount of data they collect — and learn from — over time.

As the chatbot has more conversations with site visitors, it evolves and tweaks its responses. If the AI’s response is incorrect, and then the agent changes the answer to better suit the customer’s needs, it will learn from the exchange and will answer more appropriately the next time it’s asked a similar question.

This deep learning model works in tandem with the specific domain knowledge that the chatbot now has. Now able to understand the context of the questions asked, the chatbot can give meaningful and appropriate answers.

And with NLU, the customer no longer has to intentionally speak like they’re interacting with a machine. Users of NLU-enhanced chatbots can converse via chat as they would with a friend. Technology also has the ability to categorize customer data by tagging parts of speech and reformatting numbers and dates so the machine can read them. It’s the chatbot’s job to understand the normal speech of the customer and take action accordingly.

NLU capabilities also allow the chatbot to understand customer preferences, feelings, and inclinations, resulting in in-depth insights that deepen customer relationships. For example, it could detect when a customer is becoming irate and prompt a human operator to take over the chat.

How does this help drive revenue and results?

Having capable chatbots on a website means that business owners can actually rely on them to have meaningful and contextual conversations with site visitors. These conversations result in engaged users with their expectations exceeded — ultimately increasing the likelihood of them converting into buyers.

By having knowledge of the site, accuracy is significantly increased. For example, for a retail site, a chatbot would have detailed knowledge of the items being sold on the site and can recommend appropriate products that go well with what they are looking for.

For example, ‘RecipeBots’ have recently been developed that answer questions and recommend recipes that are best suited for the user’s needs. Once the user selects a recipe, the RecipeBot adds the ingredients into a shopping cart with a single click. With this one click, the user can have all of the required products shipped to their home for them to start making their dish of choice. Businesses hosting the RecipeBot increase sales by helping users find and buy the ingredients, while site-users benefit from significantly decreasing the time and effort it takes to buy everything they need for a certain recipe.

Satisfied customers mean loyal customers, which mean higher revenues. Add to this the money saved on HR costs by employing AI to deal with what would usually go through human customer service representatives. And by taking care of repetitive tasks, AI chatbots enable human employees to spend more of their time on more interesting and stimulating work.

How can chatbots help a business run smarter?

So, for companies that are keen to deepen customer relationships and increase revenue, AI and NLU-enhanced chatbots seem to be a no-brainer. In fact, around 40 percent of large businesses have already implemented or are in the process of implementing intelligent assistants or AI chatbots, compared to 25–27 percent of SMEs.

The possibilities for chatbots to drive other, unexpected benefits for a business is high. Richard Socher, Salesforce’s chief scientist, recently commented that we can expect to see chatbots “proposing strategy and tactics for overcoming business problems.”

Not only will future chatbots be able to help marketing teams craft messages based on the understanding of the language that’s been successful in the past, but they could also develop capabilities to analyze the sentiment of the conversations they’re having, and allocate resources accordingly.

Clearly, AI and NLU-enhanced chatbots are the paths forward for businesses of all sizes that want to build customer relationships and drive results. By leaving behind traditional, rule-based chatbots and embracing new technologies, companies can make their customers feel like their time and problems are valued, gain a comprehensive understanding of consumer needs, increase product purchases, and save on HR costs in the process.

Contributed by DJ Das, Founder and CEO of ThirdEye Data.

Source: We must move away from ‘cookie-cutter’ chatbots

It’s no secret that open source software holds many benefits for businesses looking to reap the rewards of these innovative, collaborative and incidentally free efforts of developers around the world. Even major companies such as Salesforce have opened up their software for developers to test, debug and improve. Other huge players that run on open source software include Netflix and Twitter. In fact, 78 percent of companies run on open source. So, what if I told you these benefits improve exponentially for startups using OSS to launch and scale their business?

Open source software has leveled the playing field so that even small startups can leverage advanced technologies, without needing the big budget to go with it. Here are the top ways that startups seriously benefit from running on open source software solutions.

Power and Control

Startups are always are running under strict budgetary restrictions as they’re just starting out. This means executing their business model in the most economical way, and open source software undeniably allows this by relieving businesses of the need to pay the expensive initial fees of proprietary software. In fact, Linux found that its users save up to 55 percent over commercial solutions, and Barclays bank reduced its software development costs by 90 percent using Linux.

While it’s likely that down the line they may have to spend money on support, training or bespoke development needs, these choices can be made as per their business plan and actual business traction, and deployed as and when needed.

Power in the form of expanded capabilities that comes from open source software is also in large part due to the community of talented programmers who build and craft the solution, allowing for quicker troubleshooting and accelerated development. Many hands can deliver powerful results.

Innovation, Productivity and Agility

Startups need to be able to experiment freely to understand what works for their business and what doesn’t. If this need to try different options is shackled down by high costs or contractual commitment, they won’t be trying out the necessary paths or making the necessary mistakes to understand which is the best route to take — not to mention the benefit of being able to take advantage of new tools as soon as they become available, especially before other big players that have developed their own proprietary software.

Open source software allows for this kind of agility, which breeds innovation. Without the requirement to follow the dictates of a particular vendor, startups can explore all the avenues that they need to, until they know the right way for them.

For example, startups working within AI will need to identify the algorithm that works best for their particular business needs. To identify these algorithms from the library, data scientists have to perform many trial-and-error iterations. This is usually followed by a detailed data analysis and discovery phase. Only after going through these iterative cycles will the data scientist have a full understanding of which algorithm to use, how best to slice and dice the data and how best to get the expected business results.

If it wasn’t for open source software, going through these cycles would be prohibitively expensive. This is how top companies were expected to create AI algorithms in the past — with huge expenses on manpower and resources. Now, these open source technologies are mature enough to apply predictably, making AI startups easier and cheaper to build.

Open source software helps startups take huge strides in their productivity levels, too. A studyconducted by Frank Nagle, assistant professor of strategy at the Marshall School of Business at the University of Southern California, measured the impact of free open source software on productivity, especially since a growing number of businesses are increasingly using open source software as a key input.

The study found productivity either always increased or stayed neutral, as a result of open source software — the impact was never negative. For every 1 percent increase in the use of free open source software, profits can go up by $100,000. This is a “huge amount for small firms,” said Nagle.

Security and Reliability

Due to its public nature, the software’s source code is likely more reliable than anything built by independent groups of developers. And although open source software is not automatically more secure than proprietary programs, unlike closed solutions, open source software is transparent and available for inspection. So, by opening up the software and allowing anyone to fix broken code that might put it at a risk, companies can utilize the proactivity of the community to enhance the security of their programs.

Startups that are just starting out may be intimidated by the amount of open source software options that are out there, but this shouldn’t put them off — there’s a huge amount to gain by opting for open source technologies. Between saving on costs and resources, increased capacity for innovation and agility, and having the security of your systems granted by the open source software community and their dedication to great code, the options are countless for forward-thinking startups.

Source: Helping the Little Guy: The Power of Open Source Software for Startups

In this special guest feature, Dj Das, Founder & CEO of ThirdEye Data, believes that the weblogs on your website probably hold more importance than you realize. They record the intricate details of your site visitors, such as browsing behavior, clicks, and actions, and with this data come vital insights into how the server is responding, visitor actions before conversion, detecting fraudulent activities in real time, and predicting failure probabilities of the hardware infrastructure. So, outliers that skew this information are really not ideal, as they can lead to some seriously misleading insights. This is where Outlier Detection comes in. Applying Machine Learning techniques can enable you to make use of this valuable data, which can include hidden insights into your website while disposing of useless data. Dj Das founded ThirdEye in 2010 because he believed that the world is ready for a 100% data-focused products & consulting services in the Big Data and Data Sciences marketplace. He brings in strategy, execution & operational values to ThirdEye. Dj is an avid Big Data evangelist, runs & operates a meetup group named “Big Data Cloud” which is a not-for-profit organization for evangelizing around Big Data & Cloud technologies.

Fraudulent activities, hacking attempts, and operations disruptions all sound like problems every business that’s operating online should be ready to defend against. Especially as the number of cases of fraud and cyber-attacks is on the rise — $16 billion was stolen from 15.4 million U.S. consumers in 2017. Luckily, the web server logs on your website hold a lot more information than you realize, and can give vital insights into the behavior of your site visitors, ultimately uncovering such malicious activities.

In order to keep track of these potentially debilitating issues and maintain control over your web properties, you’ll need to deploy an anomaly detection solution. Anomaly detection is a machine learning-based data mining process that identifies different types of anomalies in a given data set, and helps you understand their root causes. Applying this machine learning technique enables you to glean insights from within your web server logs, and home in on anomalies that shouldn’t be there. So, exactly what can an anomaly detection solutions uncover? And how it can make it easier than ever for you to resolve and eliminate harmful activities?

Fraudulent Activity Detection through Server Traffic and Status Code Analysis

By using the data from the web server logs, an anomaly detection solution can perform server traffic analysis, and then create a baseline for what is considered normal server traffic. So, when there are deviations to this baseline, the system flags these as anomalies. Note that there is no need to write rules to define these anomalies: the machine learning system learns to identify and detect anomalies by training itself on historical data and applying the learnings on real-time data.

On any given day, if the number of server requests, total number of unique IP addresses that made requests to the server, or number of bytes transferred from the server per second are more than the baseline, an anomaly is detected.

And so, activities such as new sites from different IP addresses making focused efforts to reach the server could indicate fraudulent activity — a classic example of a Denial-of-Service (DoS) attack. The same could also be said for a known site suddenly transferring huge amounts of data, which could indicate Cyber Espionage. You need to be aware of these anomalies sooner than later — before any real damage is done to your business’ operations.

Allowing cases of such sophisticated attacks with fraudulent intent to happen unnoticed inside your website can cost your business heavily, along with sacrificing external confidence and company morale when it’s eventually uncovered.

Endpoint Analysis Can Indicate Attempted Hacks

Businesses can prevent attempted hacks or DoS attacks by using an anomaly detection solution that creates a baseline of the most popular pages of your website, while also keeping track of web pages that have accessibility issues. Being able to detect this immediately is crucial, especially with more than half of U.S. businesses having been hacked during 2017, costing them thousands of dollars in investigating the hack and restoring or replacing hardware.

Again, deviations from this baseline are considered anomalies, and could indicate harmful activity. For example, if uncommon pages are suddenly being accessed more than common pages, your website could be facing a DoS attack, which is intended to slow the site or bring it down completely.

It should now be clear why you really can’t afford to pass up on adopting an anomaly detectionsolution for your web server logs. Once the anomalies are detected, businesses can take pertinent actions, some in real-time, to prevent any malicious activities. With its ability to determine such high-risk malicious activities, an anomaly detection solution should be a staple in the systems of every business operating online.

Source: Here’s Why your Business Needs an Anomaly Detection Solution for its Web Server Logs — and Fast

Data science has become a crucial practice for many businesses to increase efficiency and to enhance processes. It has had a profound impact on a variety of areas of society. In fact, data science has been so heavily in demand that it has been ranked the best job in the US for the past three years.

However, getting into data science might be tough, initially, especially given the technical requirements of the job, but it certainly proves to be rewarding. For any potential future data scientists, Mashable recently published an article titled “Online courses in data science that could help you snag a new job.”

The article highlights just how in demand this role is stating “Companies large and small are now playing a game of tug of war on qualified data science pros, and with the proper training and tools, you just might be one of the lucky few that employers are tripping over themselves to hire.” Following this, there is a large number of online options that can help a newbie data scientist begin their journey.

In addition to Mashable, numerous sites are declaring their certainty that time spent on data science is not time wasted. One of which is CIO Review which states “It is a known fact now that the skills and expertise of data scientists have applications far beyond the tech and business sectors. As data inevitably continues to increase in quality and volume, the value of data scientists is going to exceed expectations. There isn’t an iota of doubt that data scientists have the potential to make a difference.”

Clearly, businesses can’t fill data science positions fast enough, making it ideal for anyone with this skillset seeking employment. However, if you are on the other end of the spectrum, as a business intending to make sense of a prodigious amount of data, unable to hire a tech-savvy data scientists, there are plenty of businesses out there ready to provide this service for you.

One of which is ThirdEye Data, a Silicon-Valley based one-stop-shop for Data Engineering and Data Science services & products, founded as a garage startup by a husband-wife duo focused on servicing the world of Big Data. The firm is centred around the core belief that the world is ready for a 100% data-focused products & consulting services in the Big Data and Data Sciences marketplace. Given the demand for talent within this industry, it is fair to say the world is certainly ready for this approach.

Source: Addressing the increasing demand for data scientists and how to get involved