In the last two years, artificial intelligence (AI) has started to become a mainstream topic, but core aspects of AI, like machine learning and data science, have been around for quite some time, so the technology has had plenty of time to impact the industry already.
According to a study by Tata Consultancy Services, more than 90% of companies in the energy, high tech, telecom, retail, and automotive industries use AI today. The company researched 13 industries globally and found that more than 80% of companies use AI in some capacity.
“Beyond the IT function, artificial intelligence is most often used in customer service, sales, marketing, and finance,” states the report. “In energy, 100% of companies use AI – the only industry in which every company is using it. Of the companies that don’t use AI today, all expect to by 2020.”
Globant, a global IT services company that was established in Argentina, has been involved in data-focused projects for around seven or eight years, when the company created a Big Data studio to handle huge amounts of data and build systems to do so, providing a chance to see first-hand what impacts AI has already had.
“We have seen for a long time that there are several deployments and practical applications for artificial intelligence,” said Javier Minhondo, VP of Technology – Artificial Intelligence Studio at Globant. “Very big companies are betting a lot on an AI-focused approach. They know it has the power to disrupt every business or create new opportunities, and it’s a reality and something they can deploy now, so they are trying to take advantage of it.”
As is often the case – and a topic that we address in our corporate coaching series – Globant was restricted by non-disclosure agreements that disallow any client names to be mentioned, but was happy to share some unique case studies.
Applying AI to Education
One of Globant’s clients is an e-learning company that provides a platform for kids doing home schooling in the US. Globant is building the intelligence and machine learning algorithms, including the data infrastructure, pipelines, and techniques for processing the information.
“The machine learning algorithms are there to improve the academic performance of the children,” said Minhondo. “The approach automates the recommendations system to suggest to students what subjects to work on next. The end goal is to improve efficiency, generate more engagement, reduce attrition, and reduce churn.”
The AI, in this case, is able to determine every interaction with the platform and suggest what students should be dong next. There are many data sources to do this, such as which courses the student has taken, how long it took them to finish tests, or their quality of homework, among others.
While this could be done with lots of teachers following the data of each individual student, it’s not possible to scale with this approach. There’s also a huge probability that different teachers have different approaches, so you wouldn’t get a consistent, homogeneous experience.
Practical Infrastructure Management and Challenges
On a practical scale, most companies usually have their own data, or third party data, that was acquired from APIs, reports, social media, or other data sources. This is often compiled into a data lake with a data pipeline that Nearshore tech companies can work with to build the machine learning algorithms, or execute a data science process that analyzes the quality of the data.
“Clients might have the data sources already, but we have to build a data platform in order to store it, process it, integrate it, and do everything we need to do to build a solution,” said Minhondo. “The challenges of this are not usually technical; the data scientists are usually fighting with the quality of data, so the problem is more on the cultural side of it, such as breaking silos within big companies, which are generally not 100% clear on how to manage the data. Once we are into the technical discussions it becomes a lot easier.”
Minhondo says that Globant would only suggest implementing AI if it really made sense. If a client was trying to build a chatbot, for example, and taking the first steps toward understanding what it would look like, they would not go 100% into data science and deep learning, instead leveraging certain framework or scripts to get a better sense of how users will react to it.
“One of our customers has a data source of chat sessions composed of millions of chat sessions, which we are using deep learning and building a neural network to do natural language understanding needed for the objective of the chatbot,” said Minhondo. “Yes, we are using AI, but maybe some other customer doesn’t have the same business need, or in the same stage, so just want something simpler. Ultimately, AI is not always the answer to a problem.”
Client Resources & Market Impact
Many companies are sitting on top of a lot of data, but they don’t have the resources or capabilities to leverage AI and gain vale from that data. Others may have the resources, but will be looking to do some pilot projects or R&D and need to scale up quickly. This is where the Nearshore IT service providers come in.
The IT services market is approaching artificial intelligence with a transformative attitude, as is seen in Globant’s focus.
“We want to become an AI driven company, so every project we execute in the next few years has to somehow leverage AI,” said Minhondo. “The daily impact of this is that we have to think about opportunities to do this, representing a key change in mindset, whether you’re a developer, tester, sales, or whatever – you must always think of AI first. This then leads to an idea, which becomes an executable plan, and then a design and a product.”
In terms of talent, the company is investing in people with an AI-driven mindset, also creating a large impact internally. The AI team is savvy with data-related projects, working to train the whole company, or curating content related to AI, so the whole company, and its customers, are aware of the ongoing mission.
“As we pitch it to clients, I’ve seen that there are still many incorrect perceptions that AI is a “complete” technology, and that no more advances will happen, but this is far from the truth,” said Minhondo. “Big tech companies are taking advantage of AI and machine learning, also exposing services and tool to play around with it, but their approach is very general.”
In the end, artificial intelligence is already at a mature level in terms of the tools and data that are available right now, but it still has to develop further and will ramp up very fast in the next few years, once companies start seeing the benefits and opportunities for scaling up and reducing cost.
As an IT services client, what practical impacts have you experienced with AI adoption? As a vendor, how has the technology affected the internal operations of your business? Let us know in the comments.