AI4SME: Questions you need to ask your “AI Solution Provider”
AI4SME is one of the initiative AI Singapore is championing for our Small-Medium-Enterprises (SMEs) to accelerate their adoption of AI. One of the latest initiatives launched is the Microsoft Copilot for SMEs programme in collabroation with Enterprise Singapore and Microsoft.
As SMEs start to adopt AI, one of the most common questions I get asked is “How do I know my Solution Provider (SI) partner really knows AI?”.
SMEs who wants to understand their own AI maturity can take the AI Readiness Index (AIRI) questionnaire developed by AI Singapore. In 10-15minutes, you can know if your department or organisation is AI-Unaware, AI-Aware, AI-Ready or AI-Competent.
For those in the business, you would know that just because an SI claim they have X number of PhDs on staff as AI engineers or data scientist, does not necessarily mean the SI can actually deliver an end-to-end AI solution in the end. Several large organisations have privately shared with me that their innovation teams consisting of researchers with PhDs in AI and computer science have failed to deliver any product. My question to them was always, did you even hire AI engineers and product managers to actually create a product? Or were you only trying to build an AI model?
Hidden Technical Debt of ML Systems
An AI solution is more than just have a an accurate AI model. In fact the AI model itself is often only 20-30% of the work. Most of the work goes into the whole AI pipeline including data acquistion, cleaning, transformation, understanding the data and domain, and then building the model, training and re-train the model, test, test, test, and then deploying (or serving) the model in production 24×7 with monitoring of the model performance, including a re-training pipeline. This is best described in the paper: Hidden Technical Debt in Machine Learning Systems
Is your AI Solution Provider AI-Competent?
To help our SMEs, I have compiled a list of questions an SME can ask their AI solution provider to determine if they are indeed qualified to build and deploy an AI system. This set of questions is distilled from my work with Global Partnership on AI (GPAI) working group (Broad Adoption of AI by SMEs project) experts.
- Can you clearly explain how your AI solution will integrate into my specific business processes and industry? (This question addresses the SI’s understanding of the SME’s value chain and the practical application of their AI solution.)
- Can you provide examples of successful AI projects you’ve completed for businesses similar to mine in terms of size and industry? (This question focuses on the SI’s experience and track record in delivering AI solutions to comparable clients.)
- How do you ensure the ethical use of data and AI in your solutions, and how will you address potential biases or risks specific to my business? (This question delves into the SI’s commitment to ethical AI practices and their ability to mitigate risks related to data and AI.)
- What measures do you take to protect the security and privacy of my data throughout the development and deployment of the AI solution? (This question emphasizes the importance of data security and privacy, which are critical concerns for SMEs.)
- How do you plan to involve my team in the AI project, and what kind of training or support will you provide to ensure a smooth transition and adoption of the AI solution? (This question addresses the SI’s collaboration approach and their commitment to knowledge transfer and support for the SME’s team.)
- Can you explain in simple terms how your AI solution works and what kind of results I can realistically expect to achieve? (This question aims to gauge the SI’s ability to communicate complex AI concepts in a way that is understandable to non-technical stakeholders.)
- How do you ensure the ongoing maintenance and updates of the AI solution to keep up with changing business needs and technological advancements? (This question explores the SI’s long-term commitment to supporting the AI solution and adapting it to future requirements.)
- What is your pricing model, and are there any additional costs I should be aware of, such as for training, customization, or ongoing support? (This question addresses the financial aspect of the engagement and ensures transparency regarding the total cost of ownership.)
Conclusion
The questions above are easily understood by SMEs while still covering the essential aspects of AI maturity, experience, ethics, security, collaboration, and support. By asking these questions, SMEs can gain valuable insights into the SI’s capabilities and suitability for their specific needs.
