Artificial intelligence (AI) is set to transform workplaces, disrupting daily activities.
Despite skepticism, a 2023 study reveals people mistrust AI for misplaced reasons, often trusting it where it may falter. Led by François Candelon, a study titled "How People Can Create-and Destroy-Value with Generative AI" highlights these perceptions.
In this photo illustration, Open AI's newly released text-to-video "Sora" tool is advertised on their website on a monitor in Washington, DC, on February 16, 2024.
Human-AI Collaboration in Business
Candelon collaborated with talent from esteemed institutions like MIT, Wharton, Harvard Business School, and the University of Warwick, utilizing his consulting company's staff for the experiment. He was curious about how humans and AI could effectively collaborate to benefit businesses.
The study, involving over 750 participants, assigned real tasks, including creative product innovation, where the individuals were directed to employ the assistance of OpenAI's GPT-4. Tasks ranged from pitching shoe concepts to their superiors, formulating focus group questions, to executing successful social media rollouts.
In tasks related to creative product innovation, participants using AI showed significant improvement, with around 90% enhancing their performance in tasks involving ideation and content creation.
The study demonstrated that participants using AI reached a performance level 40% higher than those without GPT-4, particularly when they accepted the technology's output suggestions without alteration. However, in tasks requiring problem-solving skills, human participants outperformed their AI-assisted counterparts, showing that certain tasks still favor human capabilities.
Interestingly, the study revealed that participants relying on generative AI for problem-solving tasks performed 23% worse than those who didn't use the tool at all. The study described generative AI as a "double-edged sword" due to its relatively uniform output, which can reduce a group's diversity of thought by 41%.
Despite this, Candelon stressed the undeniable power and inevitability of AI, quoting a famous saying that humans won't be replaced by AI but by humans using AI.
Candelon's key takeaway from the study, as shared with The Journal, is the increasing importance of data in the workplace with the integration of generative AI. He emphasizes the need for a reassessment of workflows to optimize collaboration between humans and AI, emphasizing that the study underscores the significant role data will play in this evolving landscape.
AI's Deficiency in Decision-Making and Implications for Real-World Applications
AI is designed to aid decision-making in scenarios beyond human comprehension, but it struggles with intangible human factors like ethics and morality. While AI competes favorably with the human brain in many areas, it cannot incorporate subjective experiences, feelings, and empathy that contribute to better decision-making in our world.
The reliance on algorithms and data hinders AI from analyzing decisions comprehensively, particularly missing the reasoning behind them and falling short in terms of empathy, ethics, and morality.
Real-world incidents highlight the deficiencies in AI-based decision-making. Examples include an autonomous vehicle causing a fatal accident due to inadequate object classification, an AI-based recruiting tool favoring male candidates due to biased training data, and a healthcare chatbot suggesting suicide to a patient.
These incidents showcase AI's limitations in understanding nuanced situations, ethical considerations, and the unpredictable nature of responses. In critical areas such as banking, content creation, and product recommendations, AI's inability to grasp the complete context raises concerns about its impact on human values, relationships, and public policies.
RELATED ARTICLE: US Military Aims to Replace Human Commanders With AI to Remove Bias, Accelerate Decision Making on the Battlefield
Check out more news and information on Artificial Intelligence in Science Times.