AI and Creativity—Is It Ethical; Will It Kill Creativity?

By Guest Author, Frank J. Martinez The use of Artificial Intelligence tools such as machine learning and predictive analytics will eventually become a part of the design process, it is inevitable. At first, AI will find its way into the decision-making process forming the design brief, because decision makers always seek a competitive advantage and a principal risk reduction strategy in business processes is the reduction of uncertainty. These decision makers will falsely believe that AI will grant a competitive advantage, a short cut to consumer acceptance and profit without investing research and testing. Why focus group or explore alternative designs, when the marketplace has spoken?The danger and promise of using AI in design decision-making is that AI will bake in the biases that currently exist in design practice, theory and design education. How? The processes underlying AI will rely on the use of data collections called data sets that will be made up of existing images of works of design and, when complete, may also contain consumer purchase patterns. The result will be the “baking in” of the biases of the choices made when creating the data sets. This means that the decisions with respect to style, wealth, race, gender, sexuality and cultural identities will be incorporated in the AI decision-making process. Contrary to common belief, an AI  computer can only work with the knowledge we give it and if AI is used in any design decision-making process, and it will be, such bias will form a silent but influential part of the design brief.Without a basic understanding of the processes underlying AI methods and how those processes generate a result, we can expect design decision makers will rely on a diet of culturally blind and biased design choices. This information will shape and guide the design brief, robbing it of the ineffable humanity that underlies good design. The role of the designer and design decision maker educated in the basics of AI is to be able to recognize when AI’s pernicious effects exist and to communicate the danger of overreliance on AI in the creative process. These design leaders will guide decision makers in understanding how the human-creative element in design can only be simulated by an algorithm, it can never be replaced.In a design context, a product manager may be tempted to use artificial intelligence to ask, what are the characteristics of a successful personal care product or packaging for a food product? An algorithm trained (machine learning) using existing data sets might have the “intelligence” to suggest the optimal package configuration for a product, design color way and type fonts and a retail price point in answer to the question. Furthermore, using data sets related to past purchasing patterns, buyer geodata and credit card use histories and other population data, the algorithm may even be able to suggest the optimal launch dates for such a product and provide  consumer profiles and contacts based upon purchasing histories for such goods from Amazon, Wal-Mart, and Google searches. Nowhere in that product development workflow description do the words design or designer occur.A product or brand manager could, in theory, develop a design brief that is almost entirely devoid of an inquiry about good design and how design communicates a product’s benefits or features or the relationship of the consumer to the product. If the data set is good and the algorithm is properly trained, the artificial intelligence engine will deliver an answer that will be accurate based upon the data it examines. However, that answer will also incorporate the limitations of the data sets and any biases that were incorporated into it.  What AI cannot deliver is a new design conceptualization based upon evolving trends flowing from consumer awareness, consumer behavior, buyer weariness or cultural changes. AI cannot incorporate design sensitivity to race, cultural, and gender concerns, if it is not present in the data sets. AI cannot do this because these trends, or more accurately information derived from these trends and concerns, are not, at this time, present in existing data sets.In addition to blindness as to race, culture, economic and gender issues another significant question is whether the growing use of AI reduce the decision-making power of the designer. Will the design brief devolve into mere instructions to create design based upon a narrow set of limitations, such as requirements to use defined colors, fonts and layouts? The corollary danger is that AI may reduce the designer’s ability to influence the growth of design or exploration in design. Stated simply, will AI reduce the chance for bravery in design? These are the principal challenges of the use of AI in design, in order to meet the challenge, a designer should acquire some understanding of AI.

How Does Artificial Intelligence Work and How will it Work in Design?

Artificial intelligence is the process of using mathematics to determine a best answer to a question. Such answers are reached by reviewing features and attributes that exist in a collection of information, generally called data. Stated with increased complexity, artificial intelligence is the process of using a mathematical algorithm to find the best answer to a question based upon (a) finding those features and attributes in a data set which (b) correlate to the question in a relevant manner.Artificial intelligence’s power is derived using complex mathematical analysis (algorithm) of information converted to numerical data and by this method, teaches itself to find the best answer to a question asked by the user. Using this process, a computer, will learn how to find the best answer by repeatedly applying the algorithm to the data set.  More importantly, the machine may have learned, by automatically refining the algorithm, how to apply that process to answer a question and more astonishingly, unrelated questions, with startling accuracy. On occasion, the ability of these algorithmic explorations seems to border on prescience. The important terms here are question and features. A question, is query posed by the user such as, what type of person will find me attractive based on my personality characteristics or how likely is a consumer to buy product X based upon its packages design features and/or their purchase of product Y? Stated another way, artificial intelligence is merely the seeking of an answer (or best answer) to a question using an analysis of relevant features in the data set to find the best answer. A feature in data is a character or element of the information in the data which is used as a guide to measure the “relatedness” of information in the data to the answer. Currently, design data sets are quite small and primarily limited to search libraries such as the Minst fashion data set, Imagenet (14 million images) and a painting image data set on Reddit. The Wikipedia listing for available machine learning data sets contains listings of images, sounds, twitter entries, handwriting, news, speech, and music, among others. The website Kaggle, has a data set devoted solely to the classification of artwork. Eventually, someone will create data sets devoted solely to works of design with subsets devoted to various design disciplines. In the next few years, the number and types of data sets devoted to art and design will grow and with them the temptation to use them in design decision-making. At this time AI technologies are not yet able to create works of art and design that pass the Turing test, but one day they will. Designers need to understand that AI tools will become more commonplace and their use will find their way into design decision-making. The challenge and goal for designers is to understand the basic principles of AI and to use that knowledge to help their client understand the difference between design by formula, imitations of another’s work, and genuine design that shows the thoughtful analysis of a client’s needs, their customers and the best design solution that serves all those parties.Frank J. Martinez is a former artist, designer and Patent Examiner. Frank earned a BFA from Pratt Institute and was the Production Director at Landor Associates in New York prior to attending law school. After serving as a Design Patent Examiner at the U.S. Patent and Trademark Office and an associate at several law firms, Frank founded The Martinez Group PLLC in 2008. Frank is admitted to practice law before the courts of the State of New York and the Federal District Courts for the Eastern and Southern Districts of New York as well as the Federal District Court for the Western District of Texas (Austin).Frank is also an Adjunct Professor at The School of Visual Art where he teaches Intellectual Property Law in the MFA Designer as Author and Entrepreneur Program. Frank is also a Mentor in the SVA GroundFloor Incubator Project where he counsels Incubator participants in IP Law. Frank earned an MBA in 2017 and studied advanced management at Harvard Business School’s HBX Program and is a member of the Board of the College Art Association Committee on Intellectual Property. For the past 2 years, He has studied Python Programming, AI and Machine learning at Code Academy and Coursera.