Data science /

The state of AI

July 17 2018/10 min read

The Peltarion team was proud to sponsor the 35th International Conference on Machine Learning (ICML) held in Stockholm – our hometown – earlier this year. It was an amazing opportunity to both catch up on the latest cutting-edge research and also meet with thousands of people, many of whom visited our booth to learn about our operational AI platform.

Alongside facilitating our emotion detection game, distributing copies of our AI handbook (free PDF copy available here) and talking about the Peltarion Platform, we asked people to take part in a survey on some of the key issues we believe are foremost in the AI sphere at the moment.

First off ⁠— a bias disclaimer. The ICML conference is primarily targeted towards researchers and academics, and the people who attend are often AI experts (as our survey results affirmed). For this reason, we’ve broken down some of the results between those who said they are more interested in AI for academia or research and those for business usage.

That said, the results are still very interesting and I hope this data serves to contribute to the broader discussion about AI. At Peltarion we believe the most important question facing the industry is: How can we spread the benefits of AI to more people by getting more companies, organizations, and individuals using and implementing the technology.

Interest in AI

The vast majority of the people we polled at the conference, over 60 percent, said they were mainly interested in general academia or research. However, over 35 percent said they were primarily interested in AI for business application, which indicates interest in the field is growing and is likely to continue.

Why are you interest in AI?

Educational background

Due to the academic nature of the conference, it is not surprising that nearly half of the survey respondents hold a PhD – implying that they are advanced AI users or experts.

Interestingly, a vast majority of respondents who indicated primary interest in AI for business have a master's degree (52.9 percent). while those interested with a Ph.D. narrowed in scope to just under 35 percent. This data points to the broader trend experienced by our clients: the challenge to find AI experts in the private sector.

What is your educational background?

Obstacles in developing and deploying AI

We asked what people thought were obstacles in developing and deploying AI. We got a number of different answers, but generally, the biggest obstacles respondents mentioned were related to the complexity of current tooling, problems with data, the time it takes to run and complete AI projects and the costs and complexity of necessary AI infrastructure.

It’s interesting to note that the topics that came up least were related to costs and legal or compliance issues. Again, this is likely due to the academic nature of the conference and its attendees. Had we polled company executives who were implementing AI in their business we would expect to see different results.

Access to datasets

Over 70 percent of respondents said they used online dataset libraries, but over 60 percent agreed that it was difficult to find datasets. This shows that access to datasets remains a problem for both academia and business and that current solutions are not addressing the issue.

Do you find it hard to source datasets?

Shortage of AI experts and the impact on AI adoption

Half of the respondents agreed that the scarcity of AI experts is discouraging more people from adopting AI. From our experience, this is perhaps one of the biggest obstacles facing AI adoption in the private sector.

Many people are saying there is an AI expert shortage and that is discouraging more people from adopting AI. Do you agree?

Best solution to increase AI adoption

Those most interested in AI for business application believe that “easier tools” are the best solution to increase AI adoption. This is something that also matches what we’ve heard from many clients, past and current and it’s important to understand why this is so.

Most of the best tools available today were created to help with research rather than designed for practical business usage to solve real-world problems. This has meant that for most businesses who want to get access to the most powerful and flexible solutions they are obliged to hire AI experts - but those are in extremely short supply.

Not surprisingly, respondents primarily interested in AI academia or research believe “academic or research breakthroughs” are the best solution to increase AI adoption, but it’s worth noting that “easier tools” closely followed.

What is the best solution to increase AI adoption?

General public’s understanding of AI

These results took us by surprise. Around 35 percent of respondents said the general public’s understanding of AI was “ok,” “good” or “very good.” However, just over 50 percent said it was “poor” and almost 14 percent said it was “very poor.” Thus it’s clear we’ve still got a long way to go when it comes to educating people about AI.

How would you rate the general public's understanding of AI and what it can be used for?


When it comes to educating the general public about AI and making it easier to adopt for businesses we’ve still got a long way to go. There are many initiatives around the world encouraging people – not just those with technical or computer science backgrounds – to learn more about AI, and those efforts should be applauded and encouraged. At Peltarion we’ve created a free AI Handbook to help businesses understand the technology and what they need to do to be able to operationalize it (PDF copy available here).

We believe in the good of AI and the potential it can bring to all of humanity. But to do so we need to find ways to empower more people to be able to use, experiment and operationalize AI technology. We hope that the results of our survey will contribute to the broader debate about this important issue.

  • Dominic Robinson

    PR Manager

    Dominic Robinson is responsible for Public Relations at Peltarion. He has over 10 years of experience in advocacy, media relations, public policy and EU decision-making.

02/ More on Data science