Open-source Gen AI models will gain share on closed models in market adoption.
Even in 2023 we have seen the trend shifting towards open-source Generative AI models, as evidenced by the rise of models such as Meta’s LLaMA2 , Mistral, Yi, and Solar. A key milestone was Meta's release of LLaMA, which spurred the creation of several small large language models (sLLMs), characterized by their lower parameter count but comparable performance to larger models. These sLLMs, typically ranging from 6 billion to 10 billion parameters, offer the advantages of lower cost and higher efficiency. Mistral, by Mistral AI, stood out for its efficiency and performance across various benchmarks, including math and code generation. Solar, from Upstage, is a small language model with 10.7 billion parameters, achieving top performance in essential NLP tasks while maintaining efficiency. Finally, Yi-34B, developed by 01.AI, showcased remarkable performance in language comprehension and reasoning, aiming to be a multilingual model. Additionally, techniques like LORA are facilitating this trend by simplifying the process for developers to access, experiment with these advanced models.
What fuels the overall growth of both open-source and proprietary models is definitely AI research that moves at an astonishing speed. Just search for “chain-of-thought” on Arxiv and you’ll find at least 500 papers (at the time of writing) published on this topic. Technics like that together with multi-agent RAG , RLAIF, and Tree-of-Thoughts prompting, CoT prompting and fine-tuning will continue to transform and reimagine how we build, develop, operate, and deploy AI models and apps. We will continue seeing improvements in models’ performance and efficiency all across the board with GPT-5 becoming perhaps the biggest highlight in the space in 2024. Yet, we believe that it's the collaborative innovation, customization and transparency, as well as reduced reliance on proprietary vendors brought by the open-source nature of these models, that will bolster their adoption in 2024.
Productivity gains will drive investments
Transformative potential of Gen AI is getting recognized by the business. Leading consulting firms like PwC, EY-Parthenon, McKinsey, and Cognizant provide insights into the productivity gains anticipated from the Gen AI adoption.
PwC's survey shows that 60% of CEOs expect significant efficiency gains from GenAI, indicating a strategic shift towards this technology to enhance business operations. EY-Parthenon suggests GenAI-driven productivity could lift the economy, potentially boosting real GDP by nearly 2.5% by 2033, offering a boost worth $650 billion over the next decade. McKinsey's analysis supports these findings by highlighting the potential value of GenAI across business functions, identifying areas such as customer operations, marketing, and sales, which could account for approximately 75 percent of the total annual value from generative AI use cases. Furthermore, Cognizant's projections suggest an injection of $1 trillion annually into the US economy by GenAI by 2032, emphasizing its transformative impact across sectors. Together, these insights underscore the substantial economic and productivity gains expected from GenAI, fueling the drive towards increased investment and adoption across industries.
The high expectations of productivity gains from GenAI are driving a wave of investments, as evidenced by the investment intentions of global business leaders highlighted in the following reports and surveys. The Boston Consulting Group Report indicates that a significant majority of executives plan to increase technology investments in 2024, with a special focus on AI and GenAI. Similarly, the Netcore Cloud Report reveals that around 70% of CEOs globally are planning to invest in GenAI in the upcoming year, with expectations of notable returns within a few years. The KPMG Global Tech Report also supports this, showing that a majority of global CEOs are heavily investing in GenAI, viewing it as a crucial competitive advantage. These findings highlight GenAI's growing role as a strategic asset for long-term business growth and competitiveness, underscoring the positive outlook and commitment of business leaders towards GenAI investments.
The creation of AI agents will be standardized, outsourced and specialized
Adopting AI capabilities in the corporate sector will gain ground. General purpose tools like ChatGPT can now handle a variety of tasks from summarizing texts to planning trips and are gaining adoption as a prosumer tool. The trend for specialized agents significantly progressed with the introduction of OpenAI’s GPTs - basically a marketplace of AI agents each designed to perform a certain task.
Yet, corporate needs are often even more specific, such as aiding CFOs in managing costs or helping sales teams identify new customer segments, and typically require various different tasks to be performed with sufficient level of precision and in particular order. To enable reliable, high quality commercial AI agents, there is significant research on specialized agent logic, prompting, planning and revision, tools use by agents, and fine-tuning models.
We expect a dynamic market where foundational models are widely leveraged, but having robust agent frameworks for planning and control flow, data integration, and model customization for specific sectors like healthcare, finance, and manufacturing becomes increasingly important.
Gen AI will gain ground in interpreting structured data and analytics
The integration of Gen AI in analytics and structured data management is progressively gaining momentum as an important area for AI agents. The shift focuses on developing agents that are not just vast in their knowledge base but also well-suited to handle and interpret structured data in various industries and interact with users to clarify requirements, explain data sets and iteratively provide the insights needed to make decisions. While AI tools like ChatGPT demonstrate a broad spectrum of capabilities, from language processing to complex problem solving, an important application for Gen AI lies in its proficiency in specialized analytics tasks.
This specialization is crucial for applications that require precise data interpretation, such as financial forecasting, supply chain optimization, and customer behavior analysis. The trend indicates a growing market for Gen AI solutions tailored to specific sectors, where the ability to analyze and derive insights from structured datasets is paramount. Consequently, companies might increasingly seek to integrate these specialized AI technologies, either developing in-house solutions or forming strategic partnerships with niche AI firms, to enhance their data-driven decision-making processes.
Gen AI will tackle video
Today's AI-generated videos are short, grainy, and often strange, much like the early AI-generated photos in the mid 2010s. These initial photos were basic and flawed, but quickly evolved into detailed, realistic images thanks to rapid advancements in AI technology. Similarly, AI-generated videos are poised for a transformative leap. As AI algorithms improve and computing power grows, we'll see significant enhancements in their quality and realism. A testament to this evolution is OpenAI's Sora, a cutting-edge AI model capable of producing photorealistic videos up to 60 seconds long from textual prompts, marking a notable step towards achieving high-quality, realistic AI-generated video content. This reflects the fast-paced development of AI and its potential to revolutionize video production.
100+ Chief AI Officers will be hired by Fortune 500 companies
This year, artificial intelligence has become a crucial focus in various industries, with leaders and management teams eager to understand the implications of this advanced technology for their operations. A likely trend for the coming year is the introduction of a dedicated role, the "Chief AI Officer," in large organizations. This position is expected to drive and coordinate AI strategies and projects. This development is reminiscent of the rise of cloud computing a decade ago, which saw many organizations appointing "Chief Cloud Officers" to navigate the strategic challenges and opportunities of cloud technology.
AI will fail to destroy the world
In 2024, AI will again debunk apocalyptic predictions by enhancing, not replacing, existing workflows. Gen AI will aid medical diagnoses, boost business analytics, inspire creativity in the arts, personalize education, and simplify daily life through smart technologies. This trend highlights AI's role as a collaborative tool that augments human abilities and efficiency across various sectors.
