The world’s leading publication for data science, AI, and ML professionals.

Current and Future Trends in Data Science and AI

And the impact to the corporate world

Photo by Sean Pollock on Unsplash
Photo by Sean Pollock on Unsplash

There are more mobile devices in the world than people, and every time you pick one up you interact with AI. From our smartphone cameras automatically detecting our faces when taking a selfie or unlocking our Apple phones, searching on Google for the top consulting firms in the UK or when Netflix recommending the next binge-worthy TV series to intelligent assistants like Siri and Alexa, AI-enabled systems are becoming part of our lives. These assistants utilise state-of-art AI techniques to "understand" what their user wants. But AI is not just that. As the field develops, a whole new world of opportunities emerges in front of us. This includes technologies that will help society and promote economic growth.

The pandemic has undoubtedly impacted everyone’s lives and consequently affected the corporate world. Interestingly, we are witnessing an acceleration of the digitisation of many corporations, with the introduction of new digital platforms to hold meetings to more automated back-office processes. The by-product of this digital transformation is data, data that is conveniently the blood supply for AI solutions.

Data Science and AI impact on HR function

With the near over-night shift in remote working for a large swath of the population, the nature in which business connect with their customers has become far more digitally driven. Many businesses have pivoted to online customer engagement/interactions. This change in working practices, where the boundary between work and home is blurred only by a door, is unlikely to be a temporary shift, as several benefits for employees and the businesses themselves have been observed. In particular, news reported that Google is making savings of $1 billion per year as a result of working from home, and employees benefit from better work life balance and no commute. Employers are recognising that the focus should be on what you do and not where you work. This shift in mindset will allow people to apply for jobs that were geographically locked from them previously and thus, businesses will receive far more applications than what they used to. Additionally, there will be an increase inclusion and diversity to pockets of the population – economically stimulating the ‘levelling up agenda’ – here it’s that you don’t need to live in big cities any more to have access to ‘City’ jobs.

Artificial Intelligence can assist by utilising it to build models to understand the needs of the workforce of the future, i.e., identify the key drivers for recruitment and retention. It can also assist the HR function to make decisions about candidates and specifically remove bias from the selection process. In practical terms, as part of the selection process, an AI system could audit or assist the HR function in identifying the optimal candidates in relation to a specific role. Importantly, however, if AI solutions are not thoroughly evaluated and tested, a bias might be introduced by the AI system itself. This is the reason that Amazon scrapped the AI recruiting tool developed which showed to be biased against women. Also, this advanced technology holds promise in helping to "proof-read" job descriptions to make sure they are gender-neutral and appeal to a wider and more relevant audience. An example could be scanning a job description for aggressive words or phrases like "crushing it", checking for pronouns, or words with a gender connotation, thereby increasing the diverse set of candidates that apply. Similar to how Netflix utilises AI to personalise your watchlist, AI is able to create personalised learning and development paths based on experience in order to upskill the workforce based on individual strengths and areas for development.

Data Science and AI impact on Customer Service

Many businesses have gone online and so, all customer support had to be done via phone or online chats. Chatbots will become more prevalent as they also become smarter in order to offload some of the queries from human operators. This becomes even more important in servicing customers 24/7 and potentially only triaging questions to human operator that it cannot answer. This will result in happier, less anxious customers knowing that their problems can being handled at any time rather than being on hold or not getting through to anyone. For example, as Covid-19 generated a huge number of policy checks in a time of uncertainty, an international bank wanted to reduce the time employees were spending on answering queries, while at the same time standardising responses. By utilising Google Cloud Platform, a consulting firm built an intelligent chatbot addressing these challenges by performing dynamic document search and natural conversation.

Also, since our online presence and activity has markedly increased, around 60% of total transactions are now online compared to 20–30% pre-Covid, people who didn’t necessarily used to shop online are doing so – often not out of choice but rather, Covid-related lockdowns. The ‘digital window’ into the world during periods of lockdown is allowing businesses to collect more data about their customers and monitor online behaviour in order to tailor the experience the shopper, enhancing the customer journey. For instance, one can monitor user behaviour and intervene with a pop-up chatbot if and when the user has difficulties navigating to their intended page. Or this could mean tailored preferences when navigating in terms the interface seen to better suit their expectations, or even recommendations of products and services to check out while on certain pages. In short, by learning more about the customers’ behaviour and being able to anticipate any potential issues they might face in their online journey or understand more about their needs there is an opportunity to better serve them.

As we discussed, AI systems can assist human operators by deflecting customers to different self-service channels. In order to do so, data about customer behaviour in each channel should be available and create AI systems to predict what is more relevant and helpful in each scenario.

Finally, we could not only improve the understanding of the customer calls by utilising AI to channel a phone call to the right department, but also monitor the call to understand the context, resolution and satisfaction.

Data Science and AI impact on Corporate infrastructure – closing the distance between teams

As more organisations are required to enhance their digital presence and invest in that area, there will be more use of cloud service providers and their Data Science service stacks (e.g., GCP’s AI platform, AWS’s Sagemaker or Microsoft’s AML). This enables the development of more advanced machine learning models (and specifically deep learning that fuels the likes of Google Translate/Maps etc.). From a performance point of view, these deep learning approaches are proven empirically to work best with big data, where performance of traditional statistical techniques plateau.

In order to take advantage of the infrastructure and to make the most of data, teams with different skill sets will have to work closer together. Specifically, there will be a need for more and closer collaboration between data engineers, data scientists and DevOps for the delivery of impactful artificial intelligence solutions on large scale. In short, investment in cloud services and the right set of skills will unleash the potential of AI.

Al Impact on Wellbeing

Understandably there has been a significant and needed investment in tackling the pandemic. Sharing data between countries and continents has accelerated the collective understanding of the virus dynamics. The pandemic has also forced healthcare providers to accelerate the development of AI solutions. From AI applications to differentiate various types of cough (related to covid vs not) to understanding and differentiating flu from covid symptoms via automated data science system. The main aim is to increase the speed and accuracy of diagnosis that can lead to early medical intervention and save lives.

The future can involve more predictive, preventive and personalised care. This implies the collaboration between humans and machines through augmented intelligence (AI and Human intelligence). For example, understanding and anticipating reasons for wellbeing deterioration will enable early interventions.

Also, we know that a lot of NHS trusts are becoming or have become in last few year more digitally mature, embraced cloud technologies and gathered enough data to enable the use of AI solutions for their internal processes to make them faster in processing medical records and serving of patients. These systems can be used to make predictions of needs in ICUs, forecast the length of stay of patients there and optimise the space and the nurse shifts.

Similar technologies can be utilised to optimise space and passenger flow in airports taking into consideration virus transmission and policy.

Ethical and responsible AI

Bias in data can lead to unfair models. This can be testified in a recent report by Finastra that highlights the causes and impact of bias in AI models.

AI can be a force for good, but if not implemented correctly with ethical considerations, it can be socially detrimental. Thus, organisations will have to consider how they develop and deploy AI solutions with best practices to ensure fairness, responsible and ethical use of AI. This will be an area of potential concern and scrutiny as no unified guidelines or frameworks exist at present.

Sustainable AI

Given the abundance of computation power due to the capabilities of cloud providers, there will be high risk of wrongly utilising significant resources, in certain use cases, for complex models that will have marginal or no benefit at all. This will have huge impact in the environment, and thus data scientists should be careful and thoughtful of the models and approaches they take. More specifically, creating advanced language understanding models is estimated to require as much energy as a trans-American flight. Also, there is the risk of inefficiency by trying to address different problems with an overly complex approach. This is something data science teams should be considerate of and influence their choices in terms of development.

Final Thoughts

All in all, we think about the future of AI through the lens of today’s computing. Thus, we can only discuss how AI will transform the world in the short term. However, a new computing paradigm shines on the horizon, namely quantum computing, which will take AI to an unknown yet exciting journey.


For further reading please see below:

What is Artificial Intelligence (AI)?

5 Key AI problems related to Data Privacy


Related Articles