The Report of the Select Committee on AI of the SA Parliament (published in a very busy week for AI)

Wow, what a busy week in the world of Artificial Intelligence! Good, bad, and exciting ??

Let's start with the “exciting”: Loads of announcements at Microsoft Ignite 2023, with too many to mention here. One noteworthy call out in the data analytics domain is the announcement that Microsoft Fabric, a unified software as a service (SAAS) platform that brings together Power BI, Asure Synapse Analytics, and Asure Data Factory and includes multiple experiences purpose-built for specific data worker personas, is now in general availability. Whist this is not strictly speaking AI related, Microsoft have started adding Copilot for data engineers and data scientists that use notebooks as a public preview. Those personas can now open the Copilot pane and interact with data and code through natural language, or leverage ‘magic commands’ directly inside the notebook cells—all the Copilot experiences are Fabric aware, meaning it has context about the lakehouse data and more. I will be posting more on this in the coming weeks.

Now the “bad”: this was the shock announcement of Sam Altman's sudden departure from Open AI. “Mr. Altman’s departure follows a deliberative review process by the board, which concluded that he was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities” - OpenAI announces leadership transition. While the exact reason for his departure is still being debated, this has sent shockwaves through the AI market, and even Microsoft, which has invested billions into OpenAI and integrated its technology, was caught off guard.

Now the “good”, and something a lot more local right here in Australia: this is the Report of the Select Committee on Artificial Intelligence by the South Australian Parliament.

This is a chunky report, so here is my summary (and some limited running commentary which are my own views shown in italic)) based on the report tabled in the House of Assembly and published pursuant to Standing Order 346 on 14 November 2023:

Report of the Select Committee on Artificial Intelligence by the South Australian Parliament

The report describes the current state of Artificial Intelligence (AI) development, deployment, and application across various sectors, focussing on the economic, social and ethical implications for South Australia.

As I have stated in a previous article, AI ethics and frameworks are struggling to keep up with the rate of technical change, so, it is imperative to self-regulate, so reports such as these are welcome, especially as AI is integrating itself, almost invisibly, into our everyday lives (2) Reflections from the Queensland AI Summit | LinkedIn.

The report is a culmination of submissions received over a 6-week period in July and August 2023 with 46 submissions having been received. A strong emphasis was placed on the need for continual and further investment in the field to foster innovation and economic prosperity and lead to 15 recommendations that seek to foster this enthusiasm but also mitigate any risks, including those pertaining to privacy, image-based abuse (deep fake) and job losses. The committee hopes its recommendations will help South Australia keep pace with the rapidly advancing AI technology and balance AI development with ethical considerations, education, governance, and legal frameworks.

While reading the recommendations of the report, I was reminded of my summary of the (2) Reflections from the Queensland AI Summit | LinkedIn, and want to point out that the CSIRO found that, the most successful AI projects often involve multiple partners, which suggests that collaboration is going to be key. They found that one-third of the most successful AI programs require six or more partners to succeed. So even though fostering an AI enabled workforce and culture within the public sector is very important, you need a whole ecosystem including as mentioned entities such as the Australian Institute for Machine Learning, various universities, state government entities, and industry bodies like Business SA. But the role of the private sector IMO will also play an extensive role in any aspirational success.

The 15 recommendations summarised:

Recommendation 1: Develop Sovereign AI Capability

·?????? Objective: Prioritise the development of sovereign AI capabilities for use in government departments and agencies in South Australia, including in education, healthcare, law enforcement, and agriculture.

·?????? Rationale: Enhances local protections, security, and tailored solutions for specific sectors. Addresses unique challenges like health data quality and law enforcement sensitivity.

·?????? Goals: Improve productivity, mitigate biosecurity risks, ensure equitable technology use, and maintain data security.

·?????? Call to action: Collaborate with AI developers to prioritise the development of sovereign AI capabilities, host data sets and models in South Australia, address data quality and sensitivity, and ensure equitable technology use across all South Australian demographics (including vulnerable groups).

Recommendation 2: Facilitate Appropriate Reskilling of Impacted Workers

·?????? Objective: Identify workers and industries likely to be impacted by AI and assist in their transition through appropriate support and reskilling.

·?????? Rationale: AI advancements may change job roles, necessitating upskilling to handle new technologies and tasks.

·?????? Goals: Prevent job losses, reduce disruption in small businesses, and ensure a smooth transition to AI-integrated work environments.

·?????? Call to action: Identify workers and industries likely to be impacted by AI, provide practical assistance and support for transitioning these workers, and focus on upskilling and retraining workers for AI-integrated roles.

Recommendation 3: Spearhead the Training of AI Professionals in South Australia

·?????? Objective: Increase the number of AI professionals in the state, especially in the public sector.

·?????? Rationale: To keep pace with AI advancements and to cultivate a culture of innovation in AI within South Australia.

·?????? Goals: Attract and retain AI talent, develop new AI-related roles, and upskill existing workers for AI-enhanced tasks.

·?????? Call to action: Launch initiatives to significantly increase the number of AI professionals in South Australia, invest in education, research, startup creation, and industry development to attract and retain AI talent, and upskill current workers (as previously mentioned).

Running comment: I feel there needs to be a balance between public and private, and where private, prioritise SA professionals.

Recommendation 4: Invest in Public Sector AI Integration

·?????? Objective: Develop public sector capacity for AI integration and support training for employees to understand AI's role in enhancing their work.

·?????? Rationale: AI can automate tasks, improve service delivery, and increase productivity.

·?????? Goals: Streamline administrative tasks, enhance service quality, and improve overall efficiency in government departments.

·?????? Call to action: Develop the public sector's capacity for AI integration, support training programs for employees to understand AI's role in enhancing their work and see it as a collaborative ally, and focus on how AI can automate tasks, improve service delivery, and increase productivity in government departments.

Recommendation 5: Monitor Effectiveness of AI Assisted Programs

·?????? Objective: Ensure ongoing monitoring and human oversight of AI-assisted programs in government services.

·?????? Rationale: To manage risks associated with AI, such as data-driven biases and decision-making accuracy.

·?????? Goals: Maintain accountability, ensure accurate and fair AI applications, and minimise potential errors or biases in AI systems.

·?????? Call to action: Implement continuous monitoring and human oversight of AI-assisted programs, ensure human accountability in decision-making processes involving AI, regularly assess AI effectiveness, and establish internal processes for responsible AI implementation, including impact assessment, auditing, and compliance.

Running comment: This is not limited to programs in the public sector.

Recommendation 6: Develop a State Action Plan for AI Investment

·?????? Objective: Market South Australia as the premier destination for AI investment in Australia.

·?????? Rationale: To build upon the existing AI ecosystem and attract companies and research bodies for investment.

·?????? Goals: Foster a thriving AI industry, attract significant investments, and enhance South Australia’s global reputation in AI. For this to work, there needs to be significant policy work to make South Australia a 'world leading data economy' and will need to 'develop an investment policy and strategy focused on building the industrial ecosystem around world leading areas of AI research. e.g., data centres, high performance compute, telecommunication, IoT, and human machine interfaces.'

·?????? Call to action: The Department of the Premier and Cabinet should collaborate with the Department for Trade and Investment to create a strategic plan for attracting AI investment, leverage the success of institutions like the Australian Institute for Machine Learning to promote South Australia as a premier AI investment destination, and identify and build on the industrial ecosystem around areas of AI research strength in the state.

Recommendation 7: Focus on Specialised AI Expertise Areas

·?????? Objective: Identify and develop narrow, focused areas of specialised AI expertise where South Australia can excel globally.

·?????? Rationale: Competing in selected AI niches rather than broadly across all AI technologies.

·?????? Goals: Achieve global leadership in specific AI areas like agriculture, defence, and environmental applications.

·?????? Call to action: Establish Specialised AI Expertise Areas where South Australia can excel globally, create a supportive ecosystem for AI innovation and start-up growth, and partner with institutions like the Australian Institute for Machine Learning for initiatives to foster AI adoption in industry.

Running comment: This in my opinion can only be achieved through collaboration with the private sector too.

Recommendation 8: AI Technology Growth Accelerator

·?????? Objective: Establish a Growth Accelerator program to fast-track AI adoption in industrial sectors.

·?????? Rationale: Adapt the successful model of the Manufacturing Growth Accelerator for AI technology integration.

·?????? Goals: Enhance industrial productivity and innovation through targeted AI implementation.

·?????? Call to action: Replicate the model of the Manufacturing Growth Accelerator to foster AI adoption, ensure schools have guidance on using AI tools safely and effectively, and create resources focusing on AI's ethical, safe, and pedagogically beneficial use.

Recommendation 9: AI Utilisation Framework in Education

·?????? Objective: Develop a comprehensive framework guiding schools on the ethical and effective use of AI technologies.

·?????? Rationale: Ensure AI tools are used safely and productively in educational settings.

·?????? Goals: Improve educational outcomes and prepare students for an AI-integrated future.

·?????? Call to action: Develop guidelines focusing on safe, ethical AI use and its integration into teaching and learning and provide professional development for educators on AI engagement inside and outside the classroom.

Recommendation 10: Public AI Education and Awareness Programs

·?????? Objective: Implement age-appropriate AI education programs for the public.

·?????? Rationale: Increase AI literacy and understanding across all age groups.

·?????? Goals: Educate on online safety, privacy in the digital age, and discerning misinformation, tailored to each age group.

·?????? Call to action: Create age-specific programs to raise awareness of AI, focusing on online safety, data privacy, and misinformation.

Recommendation 11: AI and STEM Integration in School Curricula

·?????? Objective: Embed AI and STEM topics across all levels of education.

·?????? Rationale: Bridge the divide between STEM and humanities, making STEM accessible to all students.

·?????? Goals: Foster early interest and competence in AI and STEM, preparing students for future technological landscapes.

·?????? Call to action: Consider a pilot program for AI-related subjects, aiming to make STEM more accessible to all students.

Recommendation 12: National AI Legislative Framework

·?????? Objective: Collaborate with the Commonwealth to create a unified legal framework for AI.

·?????? Rationale: Harmonise state and national AI laws for consistency and comprehensive governance.

·?????? Goals: Establish clear, consistent AI regulations, facilitating responsible AI development and use.

·?????? Call to action: Coordinate AI Legislation Nationally by working with the Commonwealth on for consistent AI laws and collaborate with other states for a unified AI assurance framework.

Recommendation 13: Whole-of-Government AI Governance Body

·?????? Objective: Form a permanent governmental body dedicated to AI policy and governance.

·?????? Rationale: Address the extensive impact of AI across different government sectors.

·?????? Goals: Develop an AI governance framework focused on transparency, ethical use, and societal benefits.

·?????? Call to action: Create a dedicated group for AI framework implementation, focusing on accountability, safety, and ethical use, and leverage existing frameworks and research for guidance.

Recommendation 14: Legal Review for AI-Enabled Image-Based Abuse

·?????? Objective: Assess and update legal frameworks regarding AI-enabled image-based abuse.

·?????? Rationale: Tackle the emerging challenges posed by deepfakes and similar AI technologies.

·?????? Goals: Strengthen laws to better protect individuals from AI-generated image exploitation and abuse.

·?????? Call to action: Examine current laws' effectiveness in tackling AI-enabled image-based abuse and consider AIC's suggestions like watermarking AI models and enhancing safety by design.

Recommendation 15: AI Usage Framework in Government Investigations

·?????? Objective: Establish a formal procedure for the use of AI in government investigative work.

·?????? Rationale: Standardise the application of AI in legal and investigative contexts.

·?????? Goals: Ensure ethical AI use in investigations, with proper accountability and human oversight.

·?????? Call to action: Create guidelines for the ethical and responsible use of AI technologies in government investigations and ensure human oversight.

How will SA build a competitive advantage?

·?????? South Australia has gained a competitive edge in AI by investing in the Australian Institute for Machine Learning in 2017, establishing strong research and development foundations, attracting AI investment, and building a high-quality AI talent pool.

·?????? This capability is internationally recognised, leading to significant investment and job creation in the state.

·?????? However, to maintain and grow its AI capability, South Australia needs to continue prioritising AI development, given the rapid global advancement of AI and significant investments by other governments.

·?????? South Australia is strategically focusing on specialised AI areas, rather than broadly competing with major AI players like the US and China.

·?????? The state's AI-tech ecosystem is nascent but gaining international recognition in academic and industry circles.

·?????? To enhance its competitive advantage in AI, South Australia must sustain a strong focus across education, research, startup creation, and industry development.

·?????? Investing in supporting AI technologies like communications and data management infrastructure is essential to turn AI into an innovative industry for the state.

So, what are the opportunities?

I am focussing on three of the sectors discussed in the report that I am most familiar with in the SA market.

Energy and Mining:

In South Australia's energy and mining sectors, AI offers explicit opportunities:

1.???? Enhanced Exploration and Efficiency: AI's ability to process extensive data helps in more efficient and targeted exploration, reducing costs and environmental footprint. This includes advanced techniques like natural language processing and improved image analysis.

2.???? Innovative Applications and Global Edge: AI applications are revolutionising areas like oil exploration and renewable energy, placing South Australia at a competitive advantage globally. Collaborations with AIML are leading to innovative solutions in these fields.

3.???? Safety, Sustainability, and Data Leadership: AI improves mine safety, operational efficiency, and sustainability. The use of AI for remote sensing minimises environmental impact. South Australia's leadership in digital data for exploration, facilitated by the Department for Enetgy and Mining’s initiatives, allows for virtual exploration and positions the state as a leader in AI-driven geoscience.

Education:

The use of generative AI in education in South Australia presents several opportunities, as outlined by the Department for Education. These opportunities include:

1.???? Enhancing Learning and Teaching: Generative AI tools, such as chatbots, can assist in various aspects of learning by providing personalised content and targeted feedback to students. These tools can explain complex concepts in different ways, summarise information, assist in creating study programs, test knowledge, and aid in synthesising and paraphrasing information. For teachers, AI can assist in lesson planning and reduce administrative burdens, allowing them to focus on higher-value educational tasks.

2.???? Personalised Learning Experiences: AI-powered tutoring systems, which have been around for a few years, are expected to improve with advancements in generative AI. These systems can offer dialogue-based tutoring, providing written or spoken feedback, enabling more personalised learning experiences. This is particularly beneficial for students who struggle with traditional learning methods or have specific learning needs.

3.???? Streamlining Administrative Tasks: Generative AI holds the potential to automate various routine tasks in education, thus streamlining administrative processes. This technology can be utilised for a range of activities including report writing, content creation, and planning. By automating these tasks, educators and staff can devote more time to student interaction and other high-value educational activities.

While these opportunities are promising, there are also associated risks and challenges, such as ensuring the validity and ethical use of AI-generated content, addressing potential biases in AI responses, and safeguarding data privacy and security. The South Australian education sector is committed to navigating these challenges responsibly to harness the full potential of AI in education.

Agriculture:

AI offers significant opportunities for agriculture in South Australia, a key sector contributing substantially to the state's economy. The main opportunities are:

1.????? Enhancement of Agricultural Practices: AI and machine learning are increasingly integral to precision agriculture, assisting in tasks like crop yield mapping and weed identification. They are also being used in innovative applications such as forecasting milk production and aiding in soil carbon sequestration. These technologies are transforming traditional farming practices, making them more efficient and sustainable.

2.????? Biosecurity and Pest Management: AI plays a critical role in managing agricultural pests, a major concern in crop management and biosecurity. For example, AI has been used to identify snails in grain crops, which affect grain quality, and to detect fruit flies, a significant agricultural trade pest. These AI applications help maintain high production standards and quality in Australian agriculture.

3.????? Transition to Automated Agricultural Services: AI is driving the shift towards more automated farming processes, including the use of self-driving tractors and robotic harvesters. This transition will likely result in higher outputs with lower inputs, reduced wastage, and more efficient resource distribution. AI also enhances the predictive capabilities for weather forecasts, pest and disease outbreaks, and economic modelling, which are crucial for strategic agricultural planning.

The adoption of AI in agriculture also aligns with South Australia's AgTech Strategic Plan, focusing on research, collaboration, and the commercialisation of innovative technologies. This includes projects in grain classification, vineyard management, and virtual fencing for livestock. However, alongside these opportunities, there are risks such as cyberattacks, data management challenges, socio-ecological impacts, and job transformations in the agricultural sector that need to be addressed.

My final commentary, in conclusion:

While the committee's conclusions and recommendations are comprehensive and show a keen awareness of the impact of AI, their successful implementation will require detailed planning, resource allocation, and a flexible approach to adapt to the rapidly evolving nature of AI technology. There should also be a bigger focus on collaboration with the private sector.

#SAParliament #ArtificialIntelligence #Ethics #Industry #exposéAI Exposé

?

?

?