In one year, generative AI went from a niche topic to a universally hyped concept as well as the first consumer-facing AI system to get any real traction. ChatGPT, MidJourney and Lex are used by students, social media influencers and people looking for an easy carrot cake recipe on a Sunday morning.
Despite widespread adoption and promises to revolutionize the workplace, however, business applications of AI remain questionable. It has affected some industries, ranging from customer operations to marketing and design, but wider adoption has slowed down. And this has raised momentum for critics to announce that generative models have been overhyped and might lose public attention.
3 Applications of AI With the Best ROI
- Marketing optimization.
- Content funnel optimization.
- Customer support automation.
This is a misplaced sentiment. Currently, no data-backed arguments show that generative AI will dwindle. On the other hand, this technology isn’t universally adaptable, and some industries are to gain more than others. Let’s delve deeper into generative AI applications that offer the most tangible returns on investment.
Business Applications With the Fastest ROI
The main business cases for generative AI tools are in text generation for conversational purposes, design creation and data analytics. The biggest ROI here can be achieved in three areas:
- Marketing optimization: Improved A/B testing, SEO research and market analytics.
- Content funnel optimization: Scaling and personalizing sales and marketing content, including personalized outreach.
- Customer support automation: Including virtual assistants to chatbots.
Some companies have already implemented ChatGPT into their daily operations, such as chatbots, lead generation or summarizing long sales calls and marketing videos. In the upcoming couple of years, these techniques will be more actively employed in generating new web applications and mobile apps.
Code generation is another promising use case for generative AI. However, due to the lack of precision and ability to solve complex problems, the use of artificial brains in the upcoming year should stay within the limits of less demanding tasks, such as code documentation.
Still, conservative industries like banking will be able to take advantage of generative AI for legacy code conversions. Software development and banking operations will reap the biggest benefits from further generative AI development.
Manufacturing, retail and logistics should also remain within the radar of generative AI developers, especially bearing in mind that these industries already have experience with robotic automation, including AI applications.
In retail, generative tools can be further employed to monitor real-time product status and consumer preferences, simulate production scenarios, predict demand and optimize digital shelves using historical consumer data. In manufacturing, generative AI can be trained on real-time machine data to improve predictive maintenance. In logistics, it can be used for supply chain management, such as routing optimization.
Overcoming the Generative AI Trust Trap
Despite the use-cases and promise of generative AI, companies are far from exploiting its full business potentialI. Research shows that 86 percent of IT leaders believe generative AI will have a prominent role in their organizations in the near future, according to a SalesForce report. However, 60 percent of organizations rarely use it today.
This ambiguous situation is the result of a trust gap. Only 37 percent of Salesforce customers trust AI’s outputs to be as accurate as those of a human, and 20 percent say they want a human to validate AI’s decisions. The level of trust in AI outputs has fallen in comparison to 2022. Limitations, such as hallucinations of LLMs, is a very active area of research, which seems to have become the new battleground of the AI research labs.
Mistrust is one of the reasons why Gen AI adoption has recently plateaued. Many industries — legal, pharmaceutical and manufacturing — depend on accuracy and precision. It leads to a common misconception that AI tools are suitable only for repetitive, mundane tasks. However, AI’s value lies in its ability to enhance human speed and knowledge, not replace it completely.
The end goal of an AI system shouldn’t be complete precision. AI brings value by making our processes and decisions more efficient, not by always being “right.” It can augment human knowledge and intuition with better intelligence, amplifying creative tasks, especially in situations where humans have to generate creative outputs at scale, and adding speed by doing the mundane parts of complex tasks.
Nevertheless, most applications of generative AI will have to keep a human in the loop to mitigate the risks — bias, copyright infringement or hallucinations. OpenAI has already demonstrated that LLMs can be made less biased using techniques, such as reinforcement learning from human feedback, but this is not enough.
It must be noted that further generative AI adoption in business and public organizations will highly depend on ongoing legal suits. It’s likely that next year, the industry will see some progress in institutional AI regulation and case law. At least broad-level agreements on what is and isn’t proper conduct are indispensable, especially regarding the issues of data privacy, bias and AI misuse for criminal activities.
However, 2024 is unlikely to bring legally binding norms yet. Courts might take years to solve ongoing cases. The challenge is that by that point, AI tools might become so well-established in our economies and societies that there may be no way back.
What to Expect From Generative AI in 2024
Enterprise spending on generative AI services, software, and infrastructure is expected to grow from $16 billion in 2023 to $143 billion in 2027, according to IDC forecasts.
The numbers suggest that in the upcoming couple of years, we will see new major players in the field providing training services and computing resources. One of the most promising technologies is quantum computing, which could really push the limits of artificial cognition. Though its practical applications are still under development, most developers expect to see it in full deployment in the next 10 years.
Even so, to take full advantage of generative AI’s capability to enhance efficiency and streamline processes, businesses will also have to do serious in-house work starting with training their employees. Today, data analysis, prompt engineering, AI ethics and data modeling skills are often an asset of only a few employees or highly specialized teams. If companies are going to take full advantage of what AI has to offer, those skills need to become more commonplace.
Finally, to balance out the ongoing conflict between AI developers’ interests, economic utility, and the necessity to protect privacy and other fundamental human rights, well-defined responsible AI policies will have to be passed, outlining principles of fairness, transparency, explainability and risk assessment.
Addressing these issues is critical to increasing the business ROI on generative AI tools. The future of AI lies not only in the innovations of the technology itself but also in the human ability to agree on shared principles and adopt new habits necessary for the new digital era.
