Material Selection and Sustainability Research with AI
Deliberate Academy Editorial Team
Reviewed for accuracy and professional relevance
You're 4 lessons in — don't lose your progress.
Sign up free to save where you are and earn a verified certificate when you pass.
- Use AI research tools to build a first-pass comparison of material options against performance, cost, and sustainability criteria
- Apply a source-verification standard to any sustainability claim, certification, or embodied carbon figure an AI tool produces before citing it to a client
- Identify the specific hallucination risk in AI-assisted sustainability research and explain why it is a distinct professional risk from generic factual hallucination
- Build a defensible, source-backed material comparison table for a real project using AI as a drafting and organizing tool
Material and sustainability research involves comparing dozens of products against performance specs, cost, availability, and increasingly, sustainability credentials that clients and code officials both scrutinize. AI research tools like Perplexity, which returns cited sources alongside its answers, can meaningfully speed up the first pass of this research — surfacing candidate materials and organizing comparison criteria in minutes rather than hours of manufacturer-site browsing. The risk specific to this domain: AI tools can fabricate a certification name, an embodied carbon figure, or a compliance claim that sounds exactly like the real thing, and an unverified fabricated sustainability claim repeated to a client or in project documentation is a credibility and liability problem, not just an inconvenience.
Where AI Speeds Up Material Research
Generating a candidate list. Given a performance requirement ("low-VOC, fire-rated wall panel suitable for a healthcare corridor"), AI can quickly generate a list of material categories and example manufacturers to research further — a genuine time saving over starting from a blank search.
Organizing comparison criteria. AI is useful for building the structure of a comparison table: cost tier, durability, maintenance requirements, lead time, and sustainability criteria relevant to the project (recycled content, embodied carbon, indoor air quality, end-of-life recyclability). The structure is reusable across projects even as the specific products change.
Summarizing manufacturer documentation. Once you have pulled a manufacturer's technical data sheet or Environmental Product Declaration (EPD), AI can help summarize the key figures into your comparison table faster than manual extraction — this is a use of AI on a verified source document, which is safe, distinct from asking AI to state the figures from memory.
Use AI research tools that show their sources, such as Perplexity, rather than tools that answer from memory alone. A cited source lets you click through and confirm the claim against the manufacturer's actual technical data sheet or EPD before it goes anywhere near a client deliverable. A citation is a starting point for verification, not proof by itself — check that the linked source actually says what the AI claims it says.
Material Comparison for a LEED-Targeted Office Fit-Out — Commercial Interiors Studio
Context
A studio's sustainability lead was building a material comparison for a 30,000-square-foot office fit-out targeting LEED Gold certification, with a client requirement to document embodied carbon reduction across major material categories. The comparison needed to cover flooring, ceiling systems, and wall finishes, each with three to four manufacturer options and verified sustainability data.
Action
The lead used Perplexity to generate an initial candidate list of manufacturers per material category and to surface relevant EPDs and third-party certifications for each. For every sustainability claim the tool returned, she followed the citation to the manufacturer's actual EPD or certification body listing (such as a Declare label or a Cradle to Cradle certification record) and copied the verified figure into her comparison table, discarding one AI-summarized embodied carbon figure that did not match the number in the linked EPD when she checked it directly.
Outcome
The material research phase, which had previously taken roughly a week and a half of manual manufacturer research, was completed in four working days, with every sustainability figure in the final comparison table traceable to a verified EPD or certification record rather than an AI summary alone. The discrepancy she caught between the AI-summarized figure and the actual EPD became a standing rule for the team: every embodied carbon or certification claim gets checked against its primary source before it appears in a client deliverable, with no exceptions for figures that sound right.
A designer asks an AI tool which sustainability certifications a specific carpet tile product holds. The AI names a certification that sounds legitimate but does not appear anywhere on the manufacturer's website or in any third-party certification database when checked. What does this most likely indicate?
Select one answer.
Do not repeat an AI-generated sustainability claim, certification name, or embodied carbon figure to a client, in project documentation, or in a LEED or similar certification submission until you have verified it against the manufacturer's EPD, technical data sheet, or the certifying body's own records. Fabricated sustainability claims are a specific and well-documented AI hallucination pattern, and repeating one in a formal submission can expose both you and your client to credibility damage and, in a certification context, compliance risk.
Exercise
Your Task
Choose a material category from a current or recent project. Ask an AI research tool to generate a comparison of three to four manufacturer options against your performance and sustainability criteria. For every specific figure or certification claim in the response, locate the manufacturer's actual EPD, technical data sheet, or certification body listing and confirm the claim matches. Record how many of the claims you checked matched exactly, and note any that did not.
Success looks like
- Every sustainability claim or figure in your final comparison table is traceable to a verified primary source, not just the AI response
- You have identified at least one instance where the AI summary and the primary source did not fully match, even if the difference was minor
Watch out for
- Treating a confident, well-formatted AI response as sufficient evidence on its own without checking the underlying source
- Verifying only the most surprising claims and assuming familiar-sounding certifications do not need checking — familiar-sounding fabrications are exactly the hallucination pattern this lesson warns about
Your reflection
Did you complete this exercise? What did you find? (Saved locally in your browser)
Why is AI hallucination a distinct and heightened professional risk in sustainability and material research specifically, compared to more general design research?
Select one answer.
- AI research tools that cite sources, such as Perplexity, speed up the first pass of material research — generating candidate lists and organizing comparison criteria — but a citation is a starting point for verification, not proof on its own.
- AI is safe to use for summarizing a manufacturer document you have already pulled, such as an EPD or technical data sheet — this is different from asking AI to state sustainability figures from memory.
- Fabricated certifications and embodied carbon figures are a documented AI hallucination pattern in sustainability research: plausible-sounding names and numbers that do not correspond to any real certification or verified figure.
- Never repeat an AI-generated sustainability claim, certification, or embodied carbon figure to a client or in a formal certification submission until it is verified against the manufacturer's EPD, technical data sheet, or the certifying body's own records.
- Build a comparison table where every sustainability figure is traceable to a verified primary source — this is the professional standard regardless of how the research was accelerated.