2024: The Year Artificial Intelligence Went to Work

As we reflect on the evolution of artificial intelligence (AI) from wonder to practicality, it is clear that 2024 marked a pivotal year. While 2023 captivated many with the unprecedented potential of AI, the following year focused on harnessing its capabilities for useful applications.

According to 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, there has been a notable shift from merely showcasing powerful models to actively developing viable products. This change signifies a growing maturity in the AI landscape.

When ChatGPT was launched two years ago, approximately 100 million users experimented with it, finding it beneficial in some tasks while failing in others. Today, generative AI technology is increasingly embedded in various tech services, often without users even realizing it. For instance, AI-generated responses can be seen in Google search results, as well as innovative AI techniques in photo editing tools.

Last year, many businesses released advanced generative AI models that lacked clear usage pathways for consumers. However, Narayanan notes that 2024 is seeing a gradual construction of these products designed to utilize AI capabilities for actual user benefits.

Since OpenAI unveiled GPT-4 in March 2023, accompanied by similar AI language models from competing firms, the breakthrough of continually more potent systems has subsided. This has helped to realign societal expectations, transitioning the dialogue from fears of AI taking over to understanding it as a standard technology.

During quarterly earnings calls this year, tech executives frequently encountered inquiries from Wall Street analysts about the anticipated returns from heavy investments in AI research and development. Constructing AI systems for generative tools like OpenAI's ChatGPT or Google's Gemini necessitates substantial financial input into high-energy-consuming computing systems equipped with expensive AI chips. Consequently, some tech giants announced agreements this year to utilize nuclear energy to power these operations.

Goldman Sachs analyst Kash Rangan highlighted that the investments in this technology amount to hundreds of billions of dollars. However, other analysts expressed concerns that AI has yet to solve complex problems that justify its costs, raising questions about whether AI can replicate human-like capabilities despite being trained on vast amounts of historical data.

Although Rangan acknowledges that the excitement surrounding AI has tempered since the debut of ChatGPT, he remains hopeful. He believes that AI tools are proving to be significantly more productive across various fields including sales and design.

As the technology matures, employees are left wondering if AI tools will complement their work or replace them altogether. For instance, the tech company Borderless AI utilizes a Cohere AI chatbot to draft employment contracts in countries like Turkey and India, bypassing the need for external translators and lawyers.

The Screen Actors Guild, which went on strike in July, expressed fears over AI reducing job opportunities by replicating performances without consent. Such concerns about film studios' use of AI fueled protests lasting four months last year, and video game companies have since reached agreements with the union to ensure certain AI protections.

Musicians and authors have voiced similar worries. However, experts like Walid Saad, a professor of electrical and computer engineering at Virginia Tech, assert that current generative AI cannot create truly unique or completely new works. Saad explained that while we can feed AI more data, simply having information does not equate to creativity.

AI systems lack the common sense that humans possess, which Saad argues is necessary for future advancements. Good reasoning abilities are key to enhancing AI tools for consumer usefulness. Vijoy Pandey, a senior vice president at Cisco's innovation branch, Outshift, believes that the next generation of generative AI chatbots will evolve into 'agents' capable of executing more complex tasks on behalf of users.

These AI agents could answer ambiguous questions while reasoning through steps to resolve intricate issues. Pandey envisions that by 2025, this technology will develop further, with AI agents working collectively much like a team.

As for medical applications, AI tools have streamlined processes and even provided significant assistance. For instance, the Nobel Prize in Chemistry was awarded this year for work linked to AI, promising to affect new drug discovery.

Saad also pointed out that AI aids in accelerating diagnosis by quickly analyzing data, helping medical professionals determine care paths more efficiently. However, similar to other fields, it comes with the risk of entrenching inaccuracies.

OpenAI's Whisper tool has been acclaimed for its near-human level of transcription precision, yet experts have cautioned that it can generate erroneous text, underscoring the unpredictable nature of such technologies.

In the pharmaceutical realm, the intersection of AI with traditional research has shown remarkable potential. As Pandey noted, what typically takes years in drug development can be reduced to mere days through collaborative AI use.

'To me, that has been one of the most dramatic uses of AI,' Pandey concluded.

artificialintelligence, technology, productivity, partnerships, healthcare