A Glimpse into Project Garvey
Project Garvey introduces an AI programmed to simulate the role of a human client in the design field. Its methodology includes analyzing widely-liked projects on platforms such as Dribbble, Behance, and DeviantArt. The goal is for Garvey to formulate project briefs that could have plausibly inspired such popular works, much like a game of design reverse-engineering. Where designers historically relied on A/B testing to validate design decisions, Garvey extends this logic to generating creative briefs based on trends it identifies. Upon creating a brief, Garvey engages designers via automatic postings on prominent freelance websites including 99designs, Fiverr, and Freelancer. Plans for Garvey version 1.3 include the addition of a chatbot for direct outreach to designers and agencies. To maintain the illusion of human interaction, Garvey sends periodic “How’s it going?” communications through email or SMS. Initial tests showed that instant messaging platforms like Slack and Skype made Garvey’s AI nature too evident. To further disguise its identity and provide a breadth of feedback, Garvey cycles through fictional personas—including a CEO and sales head—often providing conflicting input.
The ‘Granddaughter’ Intelligence Engine
Tom Mewling, Brigmore’s Chief Technology Officer, highlights the development of their critical intelligence engine, affectionately termed “Granddaughter,” or GD. If GD disapproves of a design, Garvey withholds approval. An element of unpredictability was designed into GD to mimic the occasionally erratic nature of human clients. However, AI’s inherent departure from human norms means Garvey can display some truly bizarre behavior. Despite Garvey’s occasionally outlandish requests, and even if a designer fulfills them, GD may still reject the outcome. Garvey treats each brief as an evolving entity, adjusting as designers submit revisions and demands an unlimited number of them. During beta testing, no designer has yet seen a project through to final sign-off. This policy raises legal considerations, particularly in light of the EU’s 2018 General Data Protection Regulation (GDPR) which mandates human oversight in decision-making, a rule not mirrored elsewhere.
Refining Client-Designer Dynamics
Brigmore aims to align client demands more closely with designers’ insights. In Mewling’s view, designers often have a more accurate grasp on what gains traction online than their human clients. By replacing human clients with an AI like Garvey, designers may enjoy greater creative liberty, remolding client problems to fit their solutions. Brigmore envisions a future where AI clients are the industry standard, with the ambitious goal of ousting human client engagement by 2025. Yet, despite the revolutionary potential, Garvey’s briefs currently offer no financial compensation, predicated on the assumption that designers are primarily motivated by exposure and portfolio enhancement. Brigmore does hold out the prospect of financial rewards for designers who reach final sign-off and may receive repeat business from Garvey—especially once it becomes profitable.
Introducing Garvey Pro
With Garvey Pro, scheduled to launch in the third quarter of 2017, designers can access different tiers of service. Inexpensive options will feature lackluster clients and uninspiring briefs, whereas premium services will be tailored for high-profile clientele, reflecting the diversity of real-world design projects which often benefit well-established firms capable of hefty pitches.
The Impact of Garvey on Design
Every stakeholder in a design project plays a part in shaping the final product. Effective briefs and insightful feedback are foundational to the design life cycle. With Garvey’s capacity to manage millions of briefs and the potential to expand, it could significantly influence the design industry. As a learning AI, Garvey evolves with each new designer interaction, gradually becoming indiscernible from a genuine client. Mewling emphasizes that for many designers, the prestige of client work in their portfolio supersedes the source. Despite an overwhelming trend towards AI-driven processes, there remains a nod to the nuances of human interaction, evidenced by classic feedback one might expect from a human client, including the inevitable request to enlarge the logo.
UPDATE: It appears this year’s April Fools’ joke has slipped by you 🙂
