The A to Z's on AI for Marketers
It seems like everyone’s talking about AI/artificial intelligence these days. In the media and in most industries, there’s a lot of curiosity and a lot of concern. Some say it can do everything humans can do and some say it will be our undoing. We all know AI isn’t going away. But what many don’t realize is that it’s already been under development since the middle of the last century. So why the sudden frenzy about it? Here’s everything marketers need to know.
What is AI?
AI is already a huge part of our lives. It’s in most computer programs, household appliances, your car and even in your pocket since smartphones are heavily reliant on it. There are a lot of AI-related terms that can be confusing, so here’s a quick overview.
1. AI Terminology
a. AI – A broad term that can include just about anything that is pre-programmed to respond to simple “if/then” situations. For example, thermostats.
b. Machine Learning – Uses an algorithm to learn and refine its understanding of optimal outputs. Think about the recommendation system on a streaming service.
c. Deep Learning – Uses layers of algorithms to learn and adapt inputs in order to create a complex or unexpected output. It’s similar to a human brain and it’s why DL is often referred to as neural networks.
2. AI Categories
a. Weak or narrow AI – Performs one function (irrespective of scale), such as facial recognition systems or self-driving vehicles.
b. Strong or general AI – Learns over time and uses those learning to solve new and different problems than those it has encountered before.
c. Super AI – A significantly higher level at which AI is substantially smarter than all humans at all levels. Currently the stuff of science fiction.
It should be noted that all current existing AI can be categorized as weak or narrow. Even the current standard bearer (ChatGPT) is a narrow AI in that the only function it performs is guessing the next words in a language sequence designed to respond to human prompts.
3. AI Functions
a. Reactive Learning – Does one thing, such as play chess. Like Deep Blue, AI can learn how to play chess by examining millions of games and determining the best and most efficient moves that will consistently beat even the best human players. However, that AI is capable of nothing else.
b. Limited Memory – Acquires information that it can use to alter its responses, but it doesn’t keep the information to use as “experience” over time. An example of this would be a self-driving car in which the car knows how to drive and acquires information from its most recent trip, but it won’t retain the latter information.
c. Theory of Mind – Understands human emotion and uses that understanding as input. For instance, an AI that has achieved theory of mind could theoretically operate as a hotline respondent for someone looking to talk through emotions.
d. Self-Awareness – Achieves self-awareness (perhaps sentience) and would exhibit behaviors indicating that the AI has developed wants, needs and goals. This is similar to Super AI.
e. To date, no AI has formally reached beyond “limited memory.” However, that is a topic of debate. ChatGPT seems to have passed the Turing Test as it is often difficult to distinguish between language generated by a human vs. AI. However, some suggest that ChatGPT has passed the Sally-Anne test (a psychological test used to determine children’s understanding of how others think, feel and behave) as well – an indication that it recognizes and correctly responds to human emotion.
What’s All the Fuss About Lately?
ChatGPT, which is a very beefed-up, much smarter version of autocomplete, is acting as the proxy for the entire industry. But there are hundreds of new platforms that have launched recently and many more that will launch in the future. The term for all these AI platforms is “generative.” Generative AI takes human input and generates content – whether that’s copy, voice, images, video, music and more.
There is no shortage of doomsayers from Geoffrey Hinton (The Godfather of AI), Eliezer Yudkowsky, Elon Musk, much of Silicon Valley and more. They are concerned that, without government regulation and responsible development, AI can lead us toward the death of truth via misinformation, robotic weapons and warfare, and human job elimination on a huge scale.
While regulation and industry standards are certainly needed, there seems to be substantial reason for optimism. Historically, technology has created substantially more jobs (entirely new industries even) than have been eliminated. AI proponents such as Reid Hoffman, Gary Vee and the CEO of Open AI (the company that created ChatGPT) feel that optimism.
Similarly, in a study that examined the potential impacts of AI, Goldman Sachs showed that 60% of occupations in the present day did not exist just 80 years ago. This cycle of technological development leading to new jobs, new industries and new tech developments has been going on for over 100 years. It is, effectively, a description of our history since the industrial revolution.
Implications to Marketers
Marketing has long been on the frontlines of technology advances (think smart phones, social media platforms, search engines, TV). It’s critical that we stay ahead of the curve in relation to these platforms. There is a saying going around the industry, “AI may not replace you, but someone that uses AI will.”
These tools are being developed to create copy, video, images, illustrations, code for websites, data sets, voice and avatar representations, 3D renderings and more. They can (and likely will) impact every avenue of the marketing environment.
Are there challenges? Absolutely. These tools have learned from source material on the internet. This raises questions about ownership, copyright, confidentiality and accountability. But marketing is highly competitive and brands can’t wait for legislators and courts to determine what’s inbounds and what’s out of bounds.
These tools can make us substantially more efficient and productive. What once took hours, may take minutes. And, while it may seem as though everyone will simply be asked to do more, it may be more likely that humans can reallocate their time to higher order tasks at which we excel – such as strategy, decision making, relationship building and continuing education. Additionally, there are qualities that AI will likely never possess such as intuition, taste, ambition and social awareness.
Thanks in part to a pandemic, for marketers in the 2020s, nearly everything we do relies (heavily) on technology. It would seem odd if a majority of people thought that we’d reached the peak of those advancements. From our perspective, this is the natural next step in the upward trajectory of technology. It’s (another) something new that offers a great deal of promise to make our lives a little easier – provided we use it the right way.
At Hoffman York, AI tools are used agencywide and have been a part of the way we conduct business for a long time. For example, the advanced technology our media team uses to buy digital media has been reliant on AI for years. It helps us to competitively win inventory, smartly identify a target and understand what ad positions are safe. Our client services team has used ChatGPT to brainstorm around audience definitions and behaviors as well as competitive strategy research. And our creative team has also used visual platforms like DALL-e and Midjourney for illustration generation as a means of inspiring new directions. They also use AI technology embedded in Photoshop to help move ideas from abstract concepts to tangible assets.
What’s Next?
Wondering if your team should be utilizing AI? Dive in – don’t hold back. Getting smart on AI is an investment that will not be time wasted. Create an internal AI task force – bring together like-minded individuals in your organization to learn as a group and explore real tools that can help you reach your goals right now. Try to get everyone using (or at least experimenting with) at least one tool. Get people comfortable with the tech and let them see that it is not intimidating. Rather, it’s incredibly helpful.
Design your company’s AI policy as you go. Determine if output should be used in your end product or if it’s only for internal use. Review what tools you’ll use as a team vs. individually. Determine if AI will be used for research and input and not allowed for “answers,” “POVs” or other intellectual pursuits. And if you’d like some guidance with developing your AI policy, the team at Hoffman York can help.
Read our article with positive predictions on how AI will change the way we as marketers and creative professionals work forever.