Helping Young Creatives Navigate Their AI Impulses
Nagin Ansari provides a guide on when and when not to use AI.
☰
Humans Using AI Will Replace Humans
Twenty-two years ago, I was interning at one of Pakistan’s premier MNCs. One that was credited with the invention of brand management and as the authority on brand building via advertising. They had it down like a well-oiled machine. There were formulae to measure the amount of brand exposure gained during a TVC. The ad duration had to be 30 seconds or under. There had to be side-by-side product demo comparisons, starting with an accepted consumer belief and ending with a benefit payoff. The company repeated this for years on end and watched its market share grow.
Simple. Until the internet, followed by social media, took the world by storm. Did they re-pivot? Of course, they did. But, in the process, they overcommitted and then course-corrected and eventually settled down to it like most companies.
Adapting to new technologies is not easy, especially when you are not sure if it is going to be a fad or the real deal. Remember the Metaverse, NFTs and the ‘everything will be based on blockchain’ mantras? However, AI seems to be here to stay – even if only going by the rate of advancement and adoption. ChatGPT hit a million users in five days (the fastest ever for a major platform launch) and that is just one of the applications. New tools are coming out every day and everything is evolving at breakneck speed.
I marvel at the pace of the advancements and at how these AI tools are accelerating productivity to almost unbelievable outcomes. The ramifications could be huge – something our industry is not waking up to fast enough in my view.
Going back to my internship days. We were tasked with developing a new TVC for one of our brands – a new campaign that required insight mining before we could write the creative brief. This is what the process involved.
1. Write the research brief for multiple focus groups and interviews across the top major cities. (Four weeks.)
2. The brand team, the creative agency and the research partners attend all sessions and make copious notes.
3. The research is complete, the team locks itself up in a meeting room for three days and writes down all the insights gathered on big flip charts. (Half a week.)
4. After much debate, the top 10 to 15 insight statements that have merit in terms of being universal and potent are narrowed down and used to develop written concepts. (One week.)
5. The concepts are put through a quantitative screener with consumers to whittle them down to the top three, following which the creative brief is written. (Eight weeks.)
6. The concepts are converted into storyboards and put through a copy-testing procedure. (Ten weeks.)
7. Based on the results, one concept is selected for execution. The costing is done, a production house and a director are selected, and the commercial is shot. (Six weeks.)
8. The final edit is completed within two weeks, incorporating comments and feedback. The TVC is ready for release and sent to the different TV stations.
The market research industry has been ripe for a technology and digital revolution for quite some time, and with the advent of AI, it can leapfrog into a whole new era. Having said this, how reliable the synthetic consumer will be compared to real consumers remains to be seen.
It has taken more than 30 weeks or half a year to develop just one TVC from scratch. From insight generation to final film, this process has not changed much in the last two decades.
Here, it is important to pause and consider the extent to which AI tools have advanced since the first version of ChatGPT was released two years ago. Morphing from a text-only advanced search engine version, ChatGPT has become a tool that accepts static and dynamic visual prompts and can assume ‘personas’ and generate multiple output formats. Videos are being created from static visuals and human-like images and videos generated. You could even have mock debates and podcasts between real people and the output would not be far from reality.
Now, let’s consider a ‘plugged-in’ brand team working at a similar MNC or a national company in today’s world. They receive similar briefs from their management to develop a new campaign for an existing brand. This is how they go about it:
1. They create the target audience personas in a GPT tool by feeding it all the research they have done with that cohort over the years. They ask this synthetic consumer all the questions they need to generate insights. (One week.)
2. They carry out a digital poll on the most popular insights they have gathered. (One week.)
3. They feed the shortlisted insights to a ‘creative’ GPT persona and ask it to churn out fully formed concept cards. They keep doing this until they are satisfied with the outcome. (Two days.)
4. The same synthetic consumer is used to answer the questionnaire after being exposed to the concept card to test the purchase intent. (Two days.)
5. The winning concept is converted into a storyboard using AI tools (two days.) The animatic is AI tested and a result is generated regarding whether it will be liked by the target audience or not (one day). If it is not liked, another animatic is generated and tested and this is repeated until a satisfactory result is obtained. (Three days.)
6. The team converts the animatic into a TVC (one day). It is tweaked based on comments and feedback and the TVC is ready to be released as a link to the TV channels. At the same time, the team has generated hundreds of digital assets to be used across social media platforms. They have also deployed their AI influencer who does not need to be paid. (Five days.)
The above process takes all of four weeks. Granted, the above is on the extreme side, but the sheer pace of advancement in some of these tools leads me to believe that this reality is not far off.There are several issues to consider, beginning with the creation of the synthetic consumer profile. It can only be as good and representative of the target audience as the data that is input into the LLM. The key here is that once the self-reasoning part kicks in with these tools, this will be when the technology no longer relies on just the inputted data. It will use its reasoning to assume, predict, reason, compare and so on.
The market research industry has been ripe for a technology and digital revolution for quite some time, and with the advent of AI, it can leapfrog into a whole new era. Having said this, how reliable the synthetic consumer will be compared to real consumers remains to be seen. Some people argue that the ability of AI tools to be completely representative of human behaviour or thinking is questionable, and that it remains dependent on the quality of data fed to the large language model and the power of the neural network it operates. However, I believe this will improve up to a point where it becomes acceptable to use synthetic consumers with disclaimers. It may still be unsuitable for categories where human interaction with the product is more intimate, such as beauty categories. However, in cases where humans play a complementary role, AI-generated visuals may become acceptable faster.
In terms of the impact on our industry, the market research and creative agencies, as well as the production houses, will have to get their act together and embrace AI sooner rather than later. We need to challenge ourselves to think about how we can incorporate these tools in our everyday work to increase productivity rather than being afraid they will replace us. As the saying goes, AI will not replace humans; humans using AI will replace humans.
Sheikh Adil Hussain is Marketing Director – Hair Care, Unilever Pakistan. sheikhadil@gmail.com
Read Comments
Related Stories
It has taken more than 30 weeks or half a year to develop just one TVC from scratch. From insight generation to final film, this process has not changed much in the last two decades.
Here, it is important to pause and consider the extent to which AI tools have advanced since the first version of ChatGPT was released two years ago. Morphing from a text-only advanced search engine version, ChatGPT has become a tool that accepts static and dynamic visual prompts and can assume ‘personas’ and generate multiple output formats. Videos are being created from static visuals and human-like images and videos generated. You could even have mock debates and podcasts between real people and the output would not be far from reality.
Now, let’s consider a ‘plugged-in’ brand team working at a similar MNC or a national company in today’s world. They receive similar briefs from their management to develop a new campaign for an existing brand. This is how they go about it:
1. They create the target audience personas in a GPT tool by feeding it all the research they have done with that cohort over the years. They ask this synthetic consumer all the questions they need to generate insights. (One week.)
2. They carry out a digital poll on the most popular insights they have gathered. (One week.)
3. They feed the shortlisted insights to a ‘creative’ GPT persona and ask it to churn out fully formed concept cards. They keep doing this until they are satisfied with the outcome. (Two days.)
4. The same synthetic consumer is used to answer the questionnaire after being exposed to the concept card to test the purchase intent. (Two days.)
5. The winning concept is converted into a storyboard using AI tools (two days.) The animatic is AI tested and a result is generated regarding whether it will be liked by the target audience or not (one day). If it is not liked, another animatic is generated and tested and this is repeated until a satisfactory result is obtained. (Three days.)
6. The team converts the animatic into a TVC (one day). It is tweaked based on comments and feedback and the TVC is ready to be released as a link to the TV channels. At the same time, the team has generated hundreds of digital assets to be used across social media platforms. They have also deployed their AI influencer who does not need to be paid. (Five days.)
The above process takes all of four weeks. Granted, the above is on the extreme side, but the sheer pace of advancement in some of these tools leads me to believe that this reality is not far off.There are several issues to consider, beginning with the creation of the synthetic consumer profile. It can only be as good and representative of the target audience as the data that is input into the LLM. The key here is that once the self-reasoning part kicks in with these tools, this will be when the technology no longer relies on just the inputted data. It will use its reasoning to assume, predict, reason, compare and so on.
The market research industry has been ripe for a technology and digital revolution for quite some time, and with the advent of AI, it can leapfrog into a whole new era. Having said this, how reliable the synthetic consumer will be compared to real consumers remains to be seen. Some people argue that the ability of AI tools to be completely representative of human behaviour or thinking is questionable, and that it remains dependent on the quality of data fed to the large language model and the power of the neural network it operates. However, I believe this will improve up to a point where it becomes acceptable to use synthetic consumers with disclaimers. It may still be unsuitable for categories where human interaction with the product is more intimate, such as beauty categories. However, in cases where humans play a complementary role, AI-generated visuals may become acceptable faster.
In terms of the impact on our industry, the market research and creative agencies, as well as the production houses, will have to get their act together and embrace AI sooner rather than later. We need to challenge ourselves to think about how we can incorporate these tools in our everyday work to increase productivity rather than being afraid they will replace us. As the saying goes, AI will not replace humans; humans using AI will replace humans.
Sheikh Adil Hussain is Marketing Director – Hair Care, Unilever Pakistan. sheikhadil@gmail.com