In 2023, the focus was on the wonder of artificial intelligence, while in 2024 efforts shifted towards making AI more practical and cost-effective. Arvind Narayanan, a computer science professor at Princeton University and co-author of “AI Snake Oil: What Artificial Intelligence Can Do, What It Can’t, and How to Tell the Difference,” noted a transition from developing models to creating tangible products. The widespread integration of generative AI technology in various services, such as Google search results and photo editing tools, reflects this trend.
Previously, companies released powerful AI models without clear applications for users. However, a shift is now occurring towards developing products that effectively leverage AI capabilities to benefit people. With the release of GPT-4 by OpenAI and similar advancements by competitors, the focus is less on continuously improving model size and quality and more on practical applications of AI technology.
Despite the potential of AI tools to enhance productivity in fields like sales and design, concerns have been raised about the significant costs associated with AI development. Analysts have questioned whether AI is truly revolutionary as initially thought, given the substantial capital investment required and the technology’s current limitations. Nonetheless, there remains optimism about the incremental benefits AI tools can bring to various professions.
As AI continues to evolve, workers are contemplating whether these tools will complement or replace their roles. Companies like Borderless AI are already utilizing AI chatbots for tasks such as drafting employment contracts, raising questions about the future impact of AI on job roles.
Video game performers belonging to the Screen Actors Guild-American Federation of Television and Radio Artists, who initiated a strike in July, expressed concerns about the potential impact of AI on their job opportunities. They feared that AI could replicate their performances without permission, leading to fewer job prospects. Similar worries were raised by musicians and authors regarding the unauthorized use of their voices and works by AI technology. Despite advancements in generative AI, experts like Walid Saad from Virginia Tech emphasize that AI lacks the basic understanding and common sense that humans possess. This limitation is evident in scenarios where AI fails to demonstrate intuitive thinking, as shown in an example involving a meme about salmon swimming in a river.
Saad points out that while AI can be trained with vast amounts of data, it does not equate to true creativity or understanding of the world. He believes that the next step for AI development is to imbue it with human-like common sense. Vijoy Pandey, a senior executive at Cisco’s innovation arm, Outshift, envisions a future where AI agents act as problem-solving ensembles, collaborating to tackle complex tasks similar to how humans work in teams. He predicts that by 2025, AI chatbots will evolve into more sophisticated agents capable of reasoning and planning. Pandey anticipates a shift towards an “agentic future,” where AI agents with specialized skills work together to perform tasks and solve problems.
In addition to entertainment and technology sectors, AI has made significant strides in the field of medicine, aiding in faster diagnostics and facilitating medical research. Notably, recent AI-related scientific advancements, such as Google’s work in discovering new medicines, have been recognized with prestigious awards like the Nobel Prize in Chemistry. Saad highlights the role of AI in providing doctors with crucial data insights for improved patient care, emphasizing that while AI cannot detect diseases, it can assist in identifying potential health issues for further evaluation.
A genuine physician should be consulted to investigate. However, like in other fields, there is a potential danger of perpetuating false information. The tech giant OpenAI has promoted its AI-driven transcription tool, Whisper, as possessing nearly “human-level robustness and accuracy.” Nevertheless, experts have pointed out a significant flaw in Whisper: it is susceptible to fabricating sections of text or even complete sentences.
Pandey, a member of Cisco, mentioned that some clients in the pharmaceutical sector have observed that AI has facilitated the connection between “wet labs,” where human beings conduct physical experiments and research, and “dry labs,” where individuals analyze data and frequently utilize computers for modeling. In the realm of pharmaceutical development, this collaborative process can span several years, but with the integration of AI, the duration can be condensed to a few days.
“That, in my view, represents the most striking application,” Pandey remarked.
