Five AI challenges to impact small businesses in the coming months

Before we dive into the challenges, let’s reset on the variety of AI systems out there – spoiler alert, it’s not just LLMs. Here are a few of the best-known alternatives:

1.      Graph Neural Networks (GNNs) are used to process complex data structures such as graphs

2.      Convolutional Neural Networks (CNNs) are for image processing and to enable computer vision

3.      Transformers are an architecture model used for AI requirements beyond language processing

4.      Reinforcement Learning (RL) is a training method for AI models that maximises rewards through trial and error enabling the model itself to evolve based on experience

5.      Generative Adversarial Networks (GANs) generate new data to resemble test data so as to create additional training data

6.      New AI chipsets enable on-device (think laptop or smartphone) AI processing to avoid expensive network, energy and data centre demand.

7.      Traditional single use A models are evolving fast for specific scientific uses such as medical diagnosis and the development of new materials and medicines.

Most small businesses are not aware of these varieties of AI and cannot therefore prepare for their arrival. However, LLMs plus a host of emerging technologies have created an urgent need for governance and risk management. Here are five areas that you will need to work on now.

1. Establish a strong and foundation for your use of AI

Define and apply your core principles for AI use, understanding that these principles will inevitably trigger changes in:

·       Governance,

·       Operational Change Management,

·       Risk Management,

·       Compliance and Control, and

·       The evolution and refinement of your Business Culture.

 Think through the governance and ethical policies you will need. All aspects of your business will be influenced by your take up of AI. Change will be rapid and must be managed.  

2. AI technology will not stand still – neither will the necessary governance and controls

The evolution of AI has been characterised by rapid invention and reinvention. Large Language Models (LLMs) have been augmented with prompt engineering which has been further enhanced with the use of AI agents. And I agents will very soon be augmented by agentic management and monitoring systems. Traditional Business Process Re-Engineering is already being augmented with automated with AI models and redefined with systems of agents. Plan how you will reengineer your business processes with AI.

 3. Agentic AI will create chaos even as it creates order

The adoption of AI agents is expanding exponentially – without proper governance, management systems and monitoring in place. There is a significant risk for businesses that adopt AI without having sufficient controls in place that they will develop uncontrolled shadow IT.  We can expect exponential growth triggering massive change, new opportunities and unmanaged risks. Implement an agent catalogue or other governance mechanism for agents now.

4. Risk management is not optional

The risks implicit in uncontrolled new AI applications will demand complex risk management from the outset. Do not wait for government regulation. Do not hesitate while jurisdictional issues are resolved. Define and enact your AI risk management approach now.

5. AI competition is ‘red in tooth and claw’

AI companies are forging ahead regardless of regulation. For national regulators, data and processing sovereignty are already top of mind – but the essential legislative and regulatory infrastructure is still in development. Expect a patchwork of disjointed policies to evolve internationally, and put your own ethical, governance, risk management and control mechanisms in place now.

For a chat about AI governance, contact mick@michaelbrawnconsulting.com.au

Mobile: 0414 987 129

www.michaelbrawnconsulting.com.au