Buying a Generative AI solution, whether it is a discrete or embedded GenAI solution, isn’t like buying a CRM or ERP system. It’s a whole new ballgame, one where you can’t always see the rules, and the players (the models) can sometimes make up their own. GenAI procurement requires a fresh playbook. Let’s break down what’s changing, why it matters, and how you can stay ahead.
What is it?
GenAI procurement is about sourcing solutions built on models that generate outputs – words, images, recommendations, decisions – based on deep, often opaque machine learning. It’s different from traditional IT procurement because:
- Transparency is critical: You need to know how and with what data the AI was trained on.
- Outputs aren’t guaranteed: Models may produce inaccuracies (“hallucinations”) with real-world consequences.
- Vendors guard secrets: Many are unlikely to disclose how their models work, which raises new negotiation challenges.
- Training Data Provenance: Where did the data used to train the model come from? Is it ethically sourced, representative, and free from problematic biases?
- Ethical expectations are higher: AI fairness, bias prevention, and explainability are not optional extras.
What does it mean from a business perspective?
It’s about managing risk – if you treat GenAI like a standard software buy, you’re adding real risks:
- Legal Exposure: Unclear IP terms can leave you vulnerable to copyright or liability issues.
- Increased Risk of Hallucinations: Without proper contractual clauses, businesses may face issues with AI-generated inaccuracies.
- Intellectual Property Ownership: Clarify who owns the AI-generated outputs and underlying models.
- Brand Damage: Bad AI outputs can quickly erode trust with customers and partners.
- Output Validation Mechanisms: Establishing processes to verify the accuracy and reliability of AI outputs.
- Vendor Lock-In: Without retraining rights, you could be stuck with underperforming or non-compliant models.
- Compliance Failures: Lack of ethical guardrails could invite regulatory or legal repercussions.
What do I do with it?
Here are some steps you can take now:
- Develop Comprehensive Contracts: Include clauses addressing hallucinations, retraining rights, IP ownership, and ethical safeguards.
- Protect IP and Ethics: Define ownership rights over outputs and require vendor bias disclosures.
- Negotiate Maximum Transparency: Work with vendors to ensure there is a clear understanding of AI models and their limitations.
- Implement Explainability Requirements: Require mechanisms that allow for the interpretation of AI decisions.
- Start with a Pilot: Validate model performance and outputs before scaling. Do your functional and non-functional testing.
- Establish Validation Processes: Create robust systems to verify the accuracy of AI outputs.
- Invest in Team Education: Ensure your procurement professionals are trained on the fundamentals of GenAI, its unique risks, and the new types of questions they need to ask vendors. (NOTE: There is also huge potential in using GenAI in the Procurement process itself to realise significant time savings and cost reductions.)
- Stay Informed: Continuously educate your procurement team on the latest developments in AI technology and procurement practices.
In a world where AI can invent, imagine, and sometimes mislead, smart procurement isn’t a luxury; it’s a necessity. Organizations that adapt now will not only avoid costly mistakes but also unlock the true value of GenAI with confidence. Ready to rethink your procurement strategy? The time to start is today.
Further Reading
Procuring AI – commercial considerations checklist (RPC – Praveeta Thayalan and Paul Joukador)