In 2023, artificial intelligence captured our imagination, but in 2024, the focus shifted to making AI practical without breaking the bank. Arvind Narayanan, a computer science professor at Princeton University and co-author of the book “AI Snake Oil: What Artificial Intelligence Can Do, What It Can’t, and How to Tell The Difference,” noted a move from just creating models to developing actual products.
Two years ago, when ChatGPT was introduced, many people were eager to interact with the chatbot, finding it either incredibly helpful or disappointingly average in different tasks. Now, generative AI technology is becoming more integrated into various tech services, such as AI-generated responses in Google searches or new AI features in photo editing tools.
Companies are now focusing on creating products that can harness the capabilities of AI models more effectively, addressing the previous challenge of releasing powerful models without clear applications for users. With the release of GPT-4 by OpenAI and similar advancements from competitors, the narrative around AI has shifted from continual leaps in intelligence to a more balanced perspective of its capabilities.
The growing investment in AI research and development has prompted discussions on the return on investment, particularly in terms of the energy-intensive computing systems required for AI applications. Tech giants are exploring solutions like nuclear power to support the energy demands of these systems. Analysts have raised concerns about the cost-effectiveness and productivity of AI models, contrasting initial optimism with the current reality.
Despite these challenges, AI is showing promise in various industries, enhancing productivity in sales, design, and other professional domains. Workers are contemplating the role of AI in their jobs, questioning whether it will complement or replace their tasks as the technology evolves. Companies like Borderless AI are already utilizing AI tools, such as chatbots, to streamline processes like drafting employment contracts for workers in different regions.
Video game performers who are members of the Screen Actors Guild-American Federation of Television and Radio Artists, went on strike in July due to concerns about how artificial intelligence (AI) could impact their job opportunities. They feared that AI could be used to replicate their performances without their consent, potentially reducing or eliminating job opportunities. Similarly, concerns about how AI could be utilized by movie studios have also led to strikes by the union in the past.
Game companies have entered into agreements with the union to protect performers from potential AI misuse. Musicians and authors have shared similar concerns about AI potentially scraping their work. While AI can be trained with more data to enhance its capabilities, it still lacks the creativity and common sense that humans possess.
Walid Saad, an AI expert at Virginia Tech, emphasized that AI currently does not understand the world in the same way humans do. He highlighted the limitations of AI by mentioning an incident where an AI engine misunderstood a prompt to create an image of salmon swimming in a river.
Looking ahead, Vijoy Pandey, a senior vice president at Cisco, discussed the potential for AI agents to enhance consumer experiences. He predicts that AI agents will be able to reason and solve complex problems on behalf of users, moving towards a future where AI agents work collaboratively as a team.
Furthermore, AI has made significant contributions to the medical field, allowing for faster diagnostics and aiding doctors in providing better care for patients. This year’s Nobel Prize in chemistry recognized AI-related advancements in drug discovery led by Google.
A genuine medical professional should investigate. However, as with other spheres, there is a risk of perpetuating misinformation. The tech behemoth OpenAI has promoted its AI-powered transcription tool, Whisper, as possessing near “human-level robustness and accuracy.” Nevertheless, experts have pointed out a significant flaw in Whisper: it has a tendency to fabricate portions of text or even entire sentences. Pandey, representing Cisco, mentioned that some clients in the pharmaceutical industry have observed that AI has facilitated bridging the gap between “wet labs,” where individuals conduct physical experiments and research, and “dry labs,” where individuals analyze data and frequently utilize computers for modeling. In the realm of pharmaceutical advancement, this collaborative procedure can span several years, Pandey noted — however, with AI, the timeline can be condensed to a matter of days. “That, in my opinion, has been the most striking application,” Pandey remarked.
