Artificial intelligence (AI) and machine learning (ML) are not new concepts. Likewise, leveraging the cloud for AI/ML workloads is not particularly new; Amazon SageMaker was launched in 2017, for example. However, there is a renewed focus on services that use AI in its various forms with the current buzz around Generative AI (GenAI).

GenAI has been getting a lot of attention lately, and rightfully so. It has huge potential to change the game for how businesses and their employees operate. of statesmen Research published in 2023 showed that 35% of individuals in the technology industry had used GenAI to assist with work-related tasks.

There are use cases that can be applied to almost any industry. Adoption of GenAI-powered tools is not limited to the tech-savvy. Using the cloud for these tools lowers the barrier to entry and accelerates potential innovation.

Related: This is the secret sauce behind effective AI and ML technology

Understanding the basics

AI, ML, deep learning (DL) and GenAI? So many terms – what's the difference?

AI can be distilled into a computer program that is designed to mimic human intelligence. This doesn't have to be complex; it can be as simple as an if/else statement or decision tree. ML takes this a step further, building models that use algorithms to learn from patterns in data without being explicitly programmed.

DL models seek to reflect the same structure of the human brain, consisting of many layers of neurons, and are excellent at identifying complex patterns such as hierarchical relationships. GenAI is a subset of DL and is characterized by its ability to generate new content based on patterns learned from large data sets.

As these methods become more capable, they also become more complex. With greater complexity comes greater demand for computation and data. This is where cloud offerings become invaluable.

Cloud offers it can generally be categorized into one of three categories: Infrastructure, Platforms and Managed Services. You may also see these referred to as Infrastructure-as-a-Service (IaaS), Platform-as-a-Service (PaaS), and Software-as-a-Service (SaaS).

IaaS offerings provide the ability to have complete control over how you train, deploy, and monitor your AI solutions. At this level, custom code will typically be written and data science experience is required.

PaaS offerings still provide reasonable control and allow you to use AI without necessarily needing a detailed understanding. In this space, examples include services like Amazon Bedrock.

SaaS offerings typically solve a specific problem using AI without exposing the underlying technology. Examples here would include Amazon Rekognition for image recognition, Amazon Q Developer for increasing software engineering efficiency, or Amazon Comprehend for natural language processing.

Practical applications

Businesses all over the world are using AI and have been for years, if not decades. To illustrate the variety of use cases across industries, take a look at these three examples from Lawpath, Careful AND Nasdaq.

Related: 5 practical ways entrepreneurs can add AI to their toolkit today

Challenges and considerations

While the possibilities are many, harnessing the power of AI and ML comes with considerations. There are many industry comments regarding ethics and responsible AI – it is essential that these are properly thought through when moving an AI solution into production.

In general, as AI solutions become more complex, their explainability decreases. What this means is that it becomes more difficult for a business to understand why a given input results in a particular output. This is more problematic in some industries than others – keep that in mind when planning your use of AI. An appropriate level of explainability is a big part of using AI responsibly.

of the ethics of AI are equally important to consider. When does it not make sense to use AI? A good rule of thumb is to consider whether the decisions your model makes would be unethical or immoral if a human made the same decision. For example, if a model refused all loans to applicants who had a certain characteristic, it would be considered unethical.

Starting

So where should businesses start with AI/ML in the cloud? We've covered the basics, some examples of how other organizations have applied AI to their problems, and touched on challenges and considerations for making AI work.

The starting point on any business' road to success adoption of AI is the identification of opportunities. Look for business areas where repetitive tasks are performed, especially those where there are decision-making tasks based on data interpretation. Additionally, look at areas where people are doing manual analysis or generating text.

With opportunities identified, objectives and success criteria can be defined. These should be clear and make it easy to quantify whether this use of AI is responsible and valid.

Only after this is determined can you start building. Start small and test the concept. Of the solutions mentioned, those at the SaaS and PaaS end of the spectrum will get you up and running faster due to a smaller learning curve. However, there will be some more complex use cases where greater control is required.

When evaluating the success of a PoC exercise, be critical and don't look at it through rose-colored glasses. As much as you, your leadership, or your investors might want to use AI, if it isn't the right tool for work, then it is better not to use it. GenAI is being touted by some as the silver bullet that will solve all problems – it's not. It has great potential and will disrupt the way many industries operate, but it is not the answer to everything.

After a successful evaluation, it is time to operationalize the capability. Think here about aspects like monitoring and surveillance. How do you make sure the solution doesn't make bad predictions? What do you do if the features of the data you used to train the ML model no longer represent the real world? Building and training an AI solution is only half the story.

Related: Unlocking AI Success – insights from leading companies on the use of AI

AI and ML are established technologies and are here to stay. Harnessing them using the power of the cloud will define the businesses of tomorrow.

GenAI is at the height of its hype and we will soon see the best use cases come out of the frenzy. To find those use cases, organizations must think innovatively and experiment.

Take the lessons from this article, identify some opportunities, test the feasibility and then make it operational. There is considerable value to be realized, but it requires due care and attention.