The State of AI

How are organisations in Australia adopting AI?

A research project made as part of my Capstone Project during my Master’s Degree.
A collaboration between The School of Cybernetics and Microsoft

After decades of research and development, Artificial Intelligence (AI) is finally touching an ever-growing number of technologies in everybody’s lives, and it’s just getting started. From virtual personal assistants in our mobile phones, to fully automated manufacturing of goods, AI has become a well-stablished technology in the last years.

However, the journey of adopting AI is unique for each organisation. We asked industry representatives and subject matter experts (SME) about the challenges and opportunities of using AI in their organisations and what could be changed to increase adoption of AI by Australian organisations. They were selected to cover different perspectives including education, industry, and government. Despite their diverse backgrounds, strong themes emerged: the need to create a culture of innovation, secure executive and top management support, provide means to safeguard data, among others.
In this report we focus on the adoption of Artificial Intelligence and Machine Learning by organisations in Australia. It is a starting point and a snapshot to understand the current state of adoption, as well as drivers and challenges that organisations encounter during this process. Further research would be needed to grasp a complete understanding of adoption of AI, including other AI techniques such as Computer Vision, Natural Language Processing, Robotics, and others.

Download the full report here

Introduction

Artificial Intelligence (AI) is a collective term for computer systems that can sense their environment, think, learn, and take action in response to what they’re sensing and their objectives. Microsoft defines AI as the ability of machines to analyse images, comprehend speech, interact in natural ways, and make predictions using data, and Machine Learning (ML) as an AI technique that uses mathematical algorithms to create predictive models.
AI could contribute up to US$ 15.7 trillion to the global economy by 2030,1 and A$ 315 billion to the Australian economy by 2028.2 The economic opportunity is great.
Recent advances in algorithms, the proliferation of digital datasets and improvements in computing capabilities have come together to speed up the adoption of AI by organisations. Many early adopters are already seeing real- life benefits of adopting AI in their digital transformation journey. However, AI adoption outside of the tech sector is at an early, often experimental stage.

An integral view of AI in organisations

How are organisations adopting AI?

Across Australia, organisations are starting to realise the potential of AI to improve their processes. Successful adoption of AI is an integral part of an organisation’s digital transformation. Organisations taking a comprehensive approach to AI have been better positioned to generate major competitive advantages. Through our research we were able to understand how organisations have included AI in their processes and what challenges they had to overcome to successfully deploy AI-based solutions. We interviewed three representatives from the education, government, and industry spaces, who shared the challenges they faced during their adoption journey. We rely in a comparative analysis, contrasting their experiences and extracting value from interviews. Ultimately, these insights are helpful to build an AI-readiness model. A summary of our findings is:

  • Applications of AI are diversifying. Whilst some organisations are focusing in adopting AI in support processes, such as data analytics and knowledge extraction, others are experimenting with predictive behaviour models and automation of repetitive tasks.
  • Projects with executive or top management support are more likely to succeed and being deployed to the entire organisation. A solid business case aligned with the organisation’s strategy proved to be useful to engage top executives in the adoption journey.
  • Organisations are creating processes and policies to collect data needed to create AI solutions. Data governance, security, and privacy are topics that organisations take in consideration when developing their data and AI strategies.
  • When creating AI solutions, organisations take a comprehensive stakeholder view, from top executives to end-users, in all stages of designing, developing, and deploying an AI solution within the organisation. They also are taking into account cultural, social, and technological aspects of the solutions.
  • In order to achieve a successful adoption of AI, organisations are upskilling their people, from top executives to developers and end- users. This ensures not only that there are enough people with the correct skillsets to develop and deploy AI but also that there is a common understanding of what AI means in the organisation.
  • Usage of Out-of-the-box solutions is getting stronger. It is in part possible due to the ease with which AI solutions can be created by end- users. Citizen development poses a new challenge in terms of governance and support if organisations are to deploy citizen developments to scale.
  • COVID-19 presented an opportunity to rethink the way AI is developed in organisations, from reimagining the way organisations work to creating new ways of engaging with stakeholders. As a consequence of COVID-19 restrictions some organisations are considering innovative solutions to emerging challenges, such as predictive models for missing data and the future of certain industries, including education and healthcare.
  • Edge computing is making a debut in Australia, especially in highly specialised industries, such as mining and manufacturing. Edge computing can also be applied in remote areas with limited or null connectivity. Industries adopting Edge computing models are taking in consideration governance, security, and support of AI models and computing modules.
  • Rapid prototyping, iterations and feedback loops proved to be successful in the adoption process. Engaging involved stakeholders during the iterations also helps to solve conflicts before the solutions is deployed. Retraining the models also help to maintain fair and accurate predictions.

Going Forward

What can organisations do to improve their AI adoption process?

In the context of AI and its adoption, the term AI-readiness assess the level of preparation of an organisation to implement changes in technologies related to AI. This term includes five categories of factors that impact an organisation’s AI-readiness: Strategic Alignment, Resources, Knowledge, Culture, and Data. Based on our research, we grouped a series of recommendations for organisations wanting to adopt AI.

  1. Develop an AI Strategy: Organisations should develop an AI strategy to bring added value to the organisation. This strategy also should include risks and mitigation procedures. Lastly, executives should show their support to AI led initiatives within the organisation
  2. Allocate Resources: Organisations should strategically allocate financial budget and personnel to support AI projects. Internal developers should be aware of existing limitations of IT infrastructure and should it need to be upscaled, be compatible with the existing stack. Lastly, organisations should create governance and support capabilities for scaling AI solutions to the entire organisation
  3. Build the Knowledge: Awareness building initiatives should be aimed to all the organisation, from top executives to end-users. It’s also helpful to manage expectations of what to expect with newly created AI solutions. Organisations should also include upskilling their people in their strategic plan.
  4. Create an AI Culture: Organisations should enable collaborative work linked to the capability of AI solutions to work across different teams and areas of the organisation. Executive support can also mean creating an environment to support innovation within the organisation which can lead to adoption of AI and related technologies.
  5. Be aware of Data: To ensure a correct and standardised access to data, organisations need to work internally, as well as with external partners and providers. The success of an AI solution is tightly linked to the quality of data available, therefore constant evaluation of data sources is required. Lastly, organisations should ensure an adequate governance of data. Data governance might include data flows and lineage, data sources and provenance, consistent data definitions, and adequate metadata management.

Going Forward

What can AI developers do to help organisations adopting AI?

In most cases, a correct adoption of AI depends equally in the organisation as well as the developers of AI. Representatives interviewed shared their experiences and they agreed that the adoption of AI in their organisations was accomplished as a consequence of a working synergy between their organisations and the developer of AI, in this case, Microsoft. They also shared some of the strategies that worked for them and how can AI developers help other organisations. We have listed some of their experiences, as well as some points in which AI developers can improve.

  • Take advantage of the momentum built by the relationship with the organisation. Build an ecosystem or technology stack in which AI is adopted naturally.
  • Work together with the organisation to develop skills needed to adopt AI. Working in a synergy with the organisation can help leverage their capability and adoption of AI.
  • Develop flexible use-cases. Finding use-cases to bring added value to the organisation can be challenging. On the other hand, out-of-the-box solutions are becoming increasingly strong and flexible to rapidly be adapted and deployed.
  • Improve Australian-centred use-cases. Experts agree that more Australia- focused datasets and solutions will help the adoption of AI by Australian organisations. There is an opportunity to capture Australian diversity in AI technologies.
  • Edge computing and citizen developers can become hard to maintain. There is an increasing concern about having the right tools to maintain, support, and govern these technologies once they are scaled to the entire organisation.

Conclusion

As our analysis underlines, the challenges that Australian organisations are facing are not broadly different from the challenges that companies are encountering worldwide. This means that Australian organisations can benefit of the existing offering of AI without compromising its performance. Out-of-the box solutions have consistently improved over the last years to offer a cost-effective way of employing AI. However, for applications that require custom made AI or ML models, organisations might encounter different challenges related to accessing Australian- centred AI solutions. AI developers will have to work together with organisations to provide feasible solutions that ensure all involved actors can access the benefits of utilising AI.
Adoption of AI has the potential to generate benefits across the economy, allowing for the same product or service to be delivered for less and be better tailored for consumers’ needs, as stated by the Australian Government.8 To ensure that Australian organisations achieve the full benefit of AI and digital innovation, we will need to better understand and develop a new perspective of digital success. For many organisations this will mean accelerating their digital transformation journey, whilst for others this will mean further building the right environment for effective adoption, upskilling people, and improving top management support. Working together, AI developers and organisations can develop actions that will result in increasing the AI-readiness and the adoption of AI by organisations.

Download the full report here
Go Back