As the landscape of supply chains grows increasingly complex and data-driven, the role of artificial intelligence becomes more critical in streamlining procurement processes. A new generation of generative AI technologies can have an even bigger impact on the field.
Generative AI, exemplified by large language models (LLMs) such as OpenAI’s GPT-4 and Google’s PaLM 2, represents a breakthrough in AI. Unlike traditional models, which typically focus on recognizing patterns or making predictions based on existing data, generative AI can learn from vast amounts of data, then generate original output, such as text, images and music.
In procurement, the relevance of generative AI lies in its ability to understand and generate human-like text. These models, trained on extensive datasets, can comprehend documents, contracts and correspondences. They can draft contracts, interpret legal language and even assist with complex negotiations, tailoring their responses to the specificities of the procurement scenario.
The technical backbone of these models is their deep learning architecture. Language models are based on neural networks, inspired by the structure of neurons in the human brain. The newer generation of models are called “large” because of their huge number of parameters, equivalent to virtual neurons — for example, GPT-3 uses 175 billion parameters.
These extremely complex systems can process vast amounts of textual data and learn patterns and relationships between words, phrases and sentences. They can understand nuanced instructions and perform many advanced cognitive tasks, matching or even exceeding human performance. However, these models still have limitations, and are prone to “hallucinations,” which is the tendency to respond to questions with highly convincing, but incorrect, information.
The relevance of generative AI in procurement is multifaceted. It has the potential to improve efficiency by automating routine tasks, with a scalability and speed that can’t be matched by human workers. Moreover, its ability to instantly analyze large amounts of data and generate insights can lead to better decision-making in areas like supplier selection, risk management and market analysis. However, it’s essential to adopt models carefully, with full human oversight in critical decision-making processes.
Applications of GenAI
Following are some specific applications of generative AI in the field of procurement.
Automated document generation and contract management. Generative AI can simplify the creation of procurement documents, including requests for proposals, purchase orders and compliance documents.
In the area of contract management, generative AI can draft, review and modify contracts based on requested criteria and standards. For instance, it can identify and adjust terms to comply with legal requirements, industry standards or company policies.
Supplier evaluation and selection. Generative AI can analyze datasets, in both structured and unstructured format, to evaluate suppliers. It can consider factors like cost, quality, reliability and past performance. The result is a supplier profile that can help procurement teams make informed decisions.
Moreover, generative AI can assist in developing negotiation strategies by analyzing previous negotiation outcomes and market conditions. It can suggest optimal pricing, terms, and conditions, tailored to each supplier.
Customized procurement. Generative AI can customize strategies based on company-specific goals, industry standards and market conditions. This can be guided by examples of previous strategy documents and files, or those explaining relevant aspects of the market.
In addition, generative AI can facilitate deeper collaboration with suppliers by enabling tailored communication and partnership strategies. For example, it can assist with the creation of partner collaterals and press releases.
Training and knowledge sharing. Generative AI can create customized chatbots based on an organization’s body of knowledge, and enable employees to ask questions, receive detailed responses and learn at their own pace.
Organizations or industry alliances can collect industry best practices, market trends, regulatory changes or other information critical for procurement teams, and create chatbots or documents that provide essential information. Another possibility is generating textual procurement scenarios, allowing teams to practice and prepare for different situations.
Generative AI is as good as the data it’s trained on, sometimes leading to unpredictable, inaccurate or inconsistent results, along with perpetuation of any biases in the data. For example, if the training data shows a preference for certain suppliers, the AI might generate recommendations that unfairly favor them. Generated outputs should always be validated and cross-checked for accuracy.
One of the main challenges with generative AI in procurement is ensuring data privacy and security. Procurement involves dealing with sensitive data, such as supplier information, pricing details and contract terms. When AI is used to generate insights from this data, there can be risks of data breaches or misuse.
In addition, as generative AI learns from existing data, there may be instances where it inadvertently reveals sensitive information. For example, if it’s trained on confidential supplier contracts, the AI might generate new contracts that include sensitive details from the original. Therefore, implementing strong data privacy and security measures, such as anonymizing datasets before training the AI, is imperative.
The decision-making process in procurement is often complex and multifaceted, involving considerations of ethics, sustainability and social responsibility. Generative AI might not fully grasp these nuances, and could potentially make decisions that are ethically questionable or socially irresponsible. It’s therefore crucial to involve humans in all decision-making, and ensure that AI suggestions align with the company’s ethical standards and social responsibility goals.
Best Practices for Deploying GenAI
Following are some best practices for the deployment of generative AI in procurement.
Train procurement teams on how to use AI tools effectively. Investing in training and development is crucial. Teams should be trained not only on how to use the tools, but also how to interpret and validate the AI’s outputs.
Furthermore, the teams should be encouraged to actively participate in the development of generative AI processes and procedures. Their expertise and insights can be invaluable in fine-tuning the interaction with generative AI and enhancing its performance.
Implement effective cybersecurity measures. This includes secure data storage and transfer, strong access controls and regular security audits.
Generative AI workflows should be designed in a way that they don’t reveal sensitive information in their outputs. Techniques like differential privacy, which adds noise to the data to prevent the identification of individual data points, can be used to enhance data privacy.
Balance AI insights with human oversight. While generative AI can automate and enhance many aspects of procurement, it can’t replace human judgment and expertise. A balanced approach is needed, whereby AI insights and human oversight are combined. AI can generate insights and recommendations, but the final decisions should be made by the procurement teams, who can take into account factors that the AI might not fully understand, such as ethics, sustainability and strategic priorities.
Generative AI holds great potential for transforming procurement processes. However, to harness this potential, it’s crucial to address the associated challenges and best practices. With careful planning and execution, generative AI can indeed become a game-changer in procurement.
Gilad David Maayan is a technology writer and head of Agile SEO.
