AI Business launched the first business-focused AI event in Tokyo last week with VisionAIres, which was held at the Grand Hyatt Tokyo Roppongi Hills. It was an evening dedicated to several roundtable discussions, where over 40 attendees, spanning some of the most forward-thinking businesses in Japan – from financial services, automotive and legal to media and technology – shared ideas on how to really advance practical AI applications in business today and in the future.

We gathered some of the key takeaways from the discussions that took place during the dinner:

Workforce Transformation

With regards to transforming the workplace with artificial intelligence it appears that RPA is a good early example of how AI capabilities will influence the workforce. The focus is primarily on back-office functions, where we have seen an increase in productivity of 30%.

“Adding deep learning increases the productivity by up to 60-70%”.

To address a concern raised by many, the discussion concluded that jobs will not be augmented by AI, as major new areas of job responsibility in the coming years will include training, supervising, and servicing robots and AI.

The discussion also revealed that we are beginning to see the move of AI projects out of IT organisations and into lines of business. The data-intensive and iterative nature of AI means that it is best when closely aligned with lines of business.

Another important takeaway to mention is that teaching and training employees on how to apply AI effectively is significantly more complex than with traditional business intelligence or data analytics tools.

There is still a question of how much AI or data science expertise an enterprise needs to maintain in-house versus how much an enterprise can rely on vendors and service providers; however, over time, the tools will catch up and will reduce the need for in-house AI expertise.

AI and Ethics

It appears that privacy issues are currently the main cause for concern, but equally for opportunity, and there is a generational shift in the view of privacy, which will further affect the attitudes about AI.

The questions raised were how much people understand the data they are giving up, and when we do give up data, do we get something of equal or greater value in return?

What appeared to be a common opinion was the need for regulation in order to protect big companies, but equally, if the companies are set out with correct intentions it can actually fuel innovation rather than stifle.

“Fundamentally, AI should be something that make people’s lives better and should have the good of humankind at the top of the agenda”.

Technologies with the Biggest Opportunity

The biggest opportunities within technology in the long-term is deep learning, as it has already proven its massive potential, and as the hardware is continuously improving, so too will the breadth of application of the technology.

It is key to outsource the expertise and establish from what regions the new talent is coming from. This will certainly drive new technologies beyond what we are utilising today and so realising where to invest in new resource is key.

A key consideration is hardware, and as the GPUs are becoming increasingly powerful, more businesses will be joining this arms race, so it is not just left to the big chip manufacturers, but others who complement the ecosystem.

The trend we see now will surely continue with NLP, Machine Learning, Deep Learning, and Image Recognition being the focus technologies under the umbrella.

Application of AI in Industry Verticals

At the top of the list is life insurance and legal services as applications of AI within industry verticals, and there is a perception of great potential in the automation of reviewing processes, claims, and other manual processes. Key areas to focus investment is the hosting of data and the data centres themselves.

However, the barriers of adoption still remain today, such as data quality, MDM, Extensive use of paper/hard copy files required for regulatory purposes, data scattered across multiple siloes and legacy systems, and the difficulty in gaining a unified view.

“Oracle is a good example of using machine learning to predict failures in data centres by looking at error logs. How could this be applied across the industries?”

Practical Applications for AI

In terms of putting AI applications into practice we have virtual and more focused sales assistants who can listen to calls, take notes, arrange personal meeting planners, and eventually evolve into independently-run sales representors.

With regards to customer service it could vary greatly geographically, as some areas are more used to automated service, such as Japan versus Europe. Security applications both on networks and more specific applications of fraud prevention and cyber hacking in government applications where AI is now being readied for military use.

Within investment we see a continuous trend for applying data to recognise the value of companies. However, leveraging more complex and predictive analytics around operational risk and longer term forecasts in a more accurate remit.