[ad_1]
While 2022 was the year of AI image generators and ChatGPT, 2023 was the year generative AI (genAI) exploded and captivated our imaginations. In 2024, the focus shifted from dazzling demonstrations to practical advancements. Models became incrementally more accurate, cost-effective, and multimodal, and enterprises began to commit to this technology.
While 60% of enterprise generative AI investments still come from innovation budgets, 40% are now coming from more permanent budgets. In a McKinsey survey of senior business leaders, more than 75% of those who deployed genAI tools at scale said those systems have met or exceeded expectations. And, compared to 2023, more CXOs are reporting that their function has rolled out the technology for at least one use case (from 4% to 22%).
Generative AI Advancements in 2024
Despite ongoing experimentation, there have been some clear use cases already driving value. For example, code copilots saw an adoption of 51%, and 77% of marketers reported that AI is helping them create more personalized content. Users of AlphaSense’s Generative Search reported saving 20 to 50 hours per month. This year, generative AI made strides in several key areas:
Multimodal Capabilities
One of the most exciting developments in genAI has been the evolution of multimodal models, which can process and generate multiple types of content simultaneously. From text-to-image generation to text-based video creation, these advancements have opened up new possibilities for applications in fields such as content creation, customer support, and education.
In March 2024, Anthropic released Claude 3, a more accurate, multimodal model that understands both text and photo inputs. In May, OpenAI released GPT-4o, which combined text, vision, and audio.
Improved Accuracy and Cost Efficiency
Generative AI models have become more reliable and affordable, with major players like OpenAI, Google, and Anthropic refining their algorithms to improve accuracy and reduce the cost of running complex models. OpenAI’s o1 models—o1-preview and o1-mini— were released in September 2024 and were designed to better tackle complex tasks that require logical reasoning. In a qualifying exam for the International Mathematics Olympiad, a high school math competition, o1 achieved 83% accuracy (OpenAI’s previous model GPT-4o achieved only 13%). And while the newer models can be more expensive—o1-preview is priced at $15 per 1 million input tokens vs GPT-4o at $5—OpenAI’s o1-mini model costs just $3 per 1 million input tokens.
Integration into Enterprise Technology
While genAI has yet to transform business processes on a large scale, we saw early steps toward enterprise integration this year. Companies in various sectors are experimenting with generative AI for tasks such as summarizing large volumes of data, generating reports, and creating personalized content. While some of these attempts did not go well—such as when Marc Benioff called Microsoft Copilot “Clippy 2.0”—these early use cases offer a glimpse into the potential of genAI as these systems improve.
Shaping the Regulatory Landscape
With the rapid rise of generative AI, governments and regulatory bodies have started to pay closer attention to its implications. In August 2024, the EU AI Act, the world’s first comprehensive AI regulation that categorizes AI applications by risk level and imposes strict requirements on high-risk AI systems, officially took effect. Meanwhile, California also proposed an AI regulation bill aimed at increasing transparency and accountability in AI systems, though it ultimately did not pass. These early regulatory efforts highlight growing concerns around data privacy, misinformation, and the potential risks of unvetted AI use.
Obstacles to Generative AI Implementation in 2024
Despite these advancements, generative AI didn’t quite achieve the widespread implementation and monetization that many anticipated in 2024. In fact, up to 85% of generative AI projects have been failing—more than double the failure rate of typical projects. Several factors have contributed to this:
Technical, Data, and Operational Challenges
A recent report highlighted that while genAI adoption grew by 17% in 2024, organizations faced significant hurdles in data preparation and quality assurance. Additionally, ensuring accuracy and reliability in diverse, real-world scenarios remains a hurdle.
Another challenge is integration with legacy systems. Many organizations must balance the promise of genAI with the realities of outdated infrastructure and siloed architectures, which can complicate deployment and limit the full potential of AI solutions.
Quantifying Business Value
While generative AI holds great promise, quantifying its impact on measurable business outcomes has proven challenging. Companies are beginning to evaluate benefits like time savings, efficiency gains, and customer engagement—all of which have been categorically improved by generative AI. Quantifying benefits beyond that—like strategic value—is still something companies are figuring out. How does one measure improved decision-making, for example?
High Implementation Costs
One of the biggest barriers to full-scale deployment of generative AI has been the high cost of implementation. Many companies did not anticipate the substantial expenses involved in running these models at scale. A report from IBM on the cost of compute for genAI cited that every executive they surveyed reported the cancellation or postponement of at least one genAI initiative due to cost this year.
2025: What’s Next for Generative AI?
Looking ahead to 2025, many of the proof-of-concept applications from 2024 are expected to transition into production. With continued improvements in accuracy, cost, and model accessibility, the coming year could see generative AI breaking out of its experimental phase and into more mainstream, strategic use. Organizations have already figured out that they don’t need to choose one model; they are now typically deploying three or more models to their AI stack. Given all this, in 2025 we can expect:
More Production-Ready Applications
As companies overcome technical and operational challenges, we’re likely to see more robust, production-ready generative AI applications across industries. Use cases that were piloted in 2024 will likely mature, with AI-powered solutions becoming integral to areas like customer service automation, knowledge management, and personalized content creation.
Strategic Wins
A McKinsey survey on how businesses are using genAI found that most use cases for this technology were applications that focused on automating tasks to save time and improve efficiency. Today’s focus on efficiency makes sense since it’s much simpler to deploy and measure. In 2025, as companies become more comfortable and ambitious with this technology, generative AI will likely go beyond delivering efficiency and time savings. We should see it being used in more strategic, innovative ways that could reshape industries.
Enhanced Quantification of AI Benefits
As more production-ready solutions are deployed, companies will develop better methods to quantify AI’s impact, moving beyond time savings to consider metrics like customer satisfaction, revenue growth, enhanced decision-making, and competitive advantage. This will help executives make informed decisions about investing in generative AI and will support its further adoption across industries.
Anticipating Needs and Autonomous Actions
In 2025, we can expect genAI systems to become significantly more proactive, moving beyond the passive “question-and-answer” model and toward intelligent anticipation of users’ needs. These systems will likely harness a deep understanding of user habits, preferences, and contexts to predict what information, assistance, or actions the user might require at any given moment. Acting as intelligent agents, these AIs could even start handling certain simple tasks autonomously, with minimal input.
In November 2024, Perplexity rolled out a feature letting Pro subscribers purchase a product without leaving its AI search engine, streamlining the research and shopping process by anticipating the user’s goal and integrating actionable next steps. While not fully autonomous, this demonstrates how genAI can proactively reduce friction in workflows by anticipating and enabling user intents. Expect to see more features of this nature in the future.
Stay Ahead of Generative AI Developments in 2025
2024 marked a pivotal year in generative AI’s evolution, defined by incremental yet meaningful progress. Advances in model capabilities have laid the groundwork for turning possibilities into production-ready solutions, even while challenges like ensuring accuracy of AI systems, data management hurdles, and high implementation costs slowed the pace of widespread adoption.
Looking to 2025, generative AI is poised to move from experimentation to real-world impact. As genAI continues to mature, businesses and innovators alike must remain flexible, capitalizing on opportunities while addressing persistent challenges, ensuring AI delivers on its transformative potential.
ABOUT THE AUTHOR
Sarah Hoffman
Director of Research, AI
Sarah Hoffman is Director of Research, AI at AlphaSense, where she explores artificial intelligence trends that will matter most to AlphaSense’s customers. Previously, Sarah was Vice President of AI and ML Research for Fidelity Investments, led FactSet’s ML and Language Technology team and worked as an Information Technology Analyst at Lehman Brothers. With a career spanning two decades in AI, ML, natural language processing, and other technologies, Sarah’s expertise has been featured in The Wall Street Journal, CNBC, VentureBeat and on Bloomberg TV, and she was named a Rising Star for WatersTechnology’s Women in Technology and Data Awards in 2018. Sarah holds a master’s degree from Columbia University in computer science with a focus on natural language processing, and a B.B.A. from Baruch College in computer information systems. Sarah is based in New York.
Read all posts written by Sarah Hoffman
[ad_2]
