AI & Art Buying: Use the tool, keep the taste human

Inspired by Artsy, Olivia Gavoyannis (Sep 15, 2025), our take for the Colonna Contemporary community/collectors.

Artificial intelligence is now part of every collector’s inbox, and yes, it’s fast. Ask a question and AI can surface artist bios, press, auction results, and rough price corridors in seconds. That speed is useful. But buying well still depends on human judgment: matching the exact series / date / size / material, reading the surface in person, and negotiating access to the right work, often the scarce work you actually want.

Work by Benjamin Wolf

Where AI shines (use it!):

• Speed & breadth. Rapid first pass on an artist’s career, shows, and ballpark pricing.

• Filter by budget. If you’re at or under a threshold (say, $10k), AI can help you avoid markets that are clearly out of range.

• Language prompts. Struggling to describe why Caravaggio hits for you? AI can help put words to “moody chiaroscuro, figurative intensity, clean surfaces”—handy for refining your search.

  
  

Where AI stumbles (watch-outs):

• Bad comps. It mixes apples and oranges, small works on paper vs. museum-level canvases—then averages the price. That’s noise, not guidance.

• Outdated bits. Sold inventory marked as “available,” old ranges, former gallery contacts—facts that were once true but aren’t now.

• No access, no diplomacy. AI doesn’t get you the call, the preview, or the blessing from an artist’s studio. Relationships do.

• Condition & fit. Screens don’t show craquelure, cockling, or how a piece breathes in your room (or a digital work on your display).

  
  

Our workflow at Colonna Contemporary (simple and calm):

1. Start broad with AI. Gather names you’re excited about.

2. Bring a shortlist to us. We verify like-for-like comps (series/date/size/material), provenance/documents, and condition risks.

3. Plan the fit. Scale, installation, and framing (or display specs for digital) so it lives beautifully at home.

4. Access & negotiation. We make the calls, request backroom options, and keep the tone professional and low-drama.

   
   

Practical example:You find a major-looking painting and see online “prices from $7k–$88k.” Without context, that range feels comforting, but it may be anchored to smaller, uncharacteristic works, not the specific series/scale you want. A better read compares only the right cohort: same series, similar year, similar size, same materials, similar exhibition weight. You’ll usually land in a defensible corridor, and you can decide whether to pay above for rarity or provenance.

How to get value from AI without getting burned:

• Add time stamps to anything you save: As of Sep 2025.

• Ask for like-for-like: “Compare only to works in the [series], circa [years], around [size], in [material].”

• Use AI to draft emails; use humans to send them. Tone and relationships matter.

• Keep a 1-page checklist on your phone (price, provenance, condition, fit) and run every listing through it in 90 seconds.

  
  

Our promise: We’ll use the tool, but keep the taste human, so you buy the thesis, not the hype.

CTA

• Want the simple plan? Reply “MAP” and I’ll send the 1-page Main Line Collector’s Guide.

• Considering a listing (fair, marketplace, another gallery, or ours)? Try our Second-Opinion Concierge for a quiet, unbiased read on price, provenance, condition, and fit, 48-hour reply.

  
  

Credit: Artsy, “pros/cons of AI for art buyers,” Olivia Gavoyannis (Sep 15, 2025).