Even as Bill Gates declares humanity has entered “The Age of AI”, and Elon Musk (rather uncharacteristically) joins a host of other technocrats in urging caution, Generative AI continues to sweep across our digital lives.

Everyday internet users and more invested techies are using ChatGPT-4 (its latest iteration) to do everything from creating basic codes to passing competitive law and business exams, to creating recipes out of items you have in your fridge, with Google’s Bard AI chatbot having just entered the fray.

Generative AI is also being used to, what else, generate images of historical figures, fictionalised speeches by current world leaders, and celebrity videos that look legitimate, all of which are created by typing a few sentences in a chat box. The potential, whether for good or ill, is mind-boggling. In this article, we look at the cons, even as a companion article celebrates the good.

“The era of AI is unquestionably here, and there will be a rush to test this ground-breaking new tool. Given the current hype, it's critical for brands to step back and thoroughly comprehend the technology before jumping in,” says Samir Asher, Co-founder and COO, Tonic Worldwide, adding that rushing into the AI world without proper planning and strategy might not produce the results one would anticipate.

Harikrishnan Pillai, CEO and co-founder, TheSmallBigIdea, believes that a key issue with AI in advertising is the issue of quick versus effective. In advertising, where stakeholders are looking for sharp solutions and not necessarily quick fixes, this can get tricky.

For instance, “AI complicates the data privacy situation. With AI comes the ability to gather and analyse massive amounts of data, but there will also be situations where one will not be able to control the sources from which the data is collected. It becomes the brand’s responsibility to be conscious and transparent about the raw data that is being manipulated.”

According to Asher, “Brands must first make sure they have the necessary data before they can begin to build their models. It is extremely important to use relevant, high-quality datasets because the training data's quality determines how successful AI will be. Secondly, brands must be conscious of any potential biases that might be incorporated into their models. Brands must be aware of any biases in their data sets and take action to resolve them because AI is only as objective as the data it is trained on,” says Asher.

Pillai agrees that the bias problem is a key one, saying, “At TheSmallBigIdea, we do periodic audits to remove any biases from the data and work on decision-making algorithms, not just data-sourcing algorithms. We merge two AI algorithms to plug into our own in order to have solid logic behind the decision.”

Dia Kirplani, Head of Strategy at Blink Digital notes that there are many stumbling blocks that brands can face if their use of AI is in untrained hands. “But the two most common ones to watch out for, which we have seen from others in the market, are inaccuracies and common messaging across different brands. That's a sure sign of AI being used poorly,” she says.

Brand safety is another factor. “Imagine that the AI places your ad on a website with offensive content just because the traffic for your target group is highest there. While AI can help deliver, the brand needs to constantly audit the quality of delivery, not just quantitatively but also qualitatively,” says Pillai, adding that there are also some ethical concerns.

“A lot of the data available might propagate or be mounted on harmful stereotypes. AI now studies it and gives an outcome. A classic example is the recent AI-generated faces of Hollywood actors that show how they would look if they were Indians,” he points out.

As AI is growing, industry insiders say there has been a major spike noticed in tech start-ups that are only offering data science and AI-powered products and services.

“Hence, in the coming days, there will be a major competition in tools and technology that may have similar features but with different pricing, data sources and other analytics, which eventually will increase the grey market, data fraud and other unethical practices,” says Danish Malik, Co-Founder and CEO of Boomlet Media Private Limited.

Here, brands need to be very selective on what technology they will be opting for AI understanding and its implementation and the credibility of their agencies providing the same will matter the most. Because if the data is extracted or stored illegally, there's a major problem or even a wrong placement may change the results upside down, which will be disastrous in case of wrong decisions being made.

For Mithun Mukherjee, ECD, Kinnect, the single most critical pitfall of AI would be something that is being warned about by global giants like Google.

“Sundar Pichai made a not-so-difficult-to-decode comment about Google Bard, which sums up the current state of the technology quite succinctly - things will go wrong. This was in reaction to misinformation that the AI gave as a part of its own promotion, post which it was pulled down,” says Mukherjee.

“Ensuring authentication of information, being able to differentiate between original vs AI-generated content and finally, doing their due diligence when using or applying AI would be some guardrails that brands would need to put into place,” says Mukherjee.