AI and GenAI hype is subsiding?
The AI/GenAI wave that began with the release of ChatGPT, giving a much wider audience access to AI, is now showing signs of a shift.
The AI/GenAI wave that began with the release of ChatGPT, giving a much wider audience access to AI, is now showing signs of a shift. While AI existed before, ChatGPT's accessible format triggered a rush to adopt GenAI, driven by FOMO, ease of use, and a lower barrier to entry.
However, after a couple of years, it seems we're seeing a recalibration. Here are some key indicators:
Slowing LLM Development: The pace of new Large Language Model releases appears to be moderating. The significant costs associated with developing these models, coupled with the challenge of achieving substantial performance gains with each iteration, are likely contributing factors.
ROI Realities: Companies that adopted AI primarily due to FOMO are now facing the reality of higher operational costs and potentially lower-than-expected returns on investment.
Chatbot Challenges: The adoption and ROI of chatbots in customer support and similar applications are facing hurdles, with exceptions in high-volume scenarios like food delivery, e-commerce, telecom, and banking.
Developer Challenges: We're seeing instances where development teams with limited AI expertise are encountering challenges in production, facing security vulnerabilities or performance and reliability issues.
Cost of Fine-Tuning: The initial belief that fine-tuning and self-hosting would be the optimal path is being tempered by the realization of the substantial costs involved, often without commensurate ROI.
Agent Development Focus: There's a growing recognition that AI agent development needs to shift from general-purpose assistants to solutions addressing specific business problems.
Despite this recalibration, GenAI will undoubtedly transform how we build solutions. It will continue to drive advancements in areas like developer productivity, business process automation, knowledge discovery, and research.
While the pace may be more measured, new models with enhanced capabilities will emerge. I believe we'll see these trends continue to shape AI development:
Integrated Functionalities: LLMs will increasingly be bundled with application-specific features such as web search, computer interaction, agent capabilities, and other tools.
AI-Native Businesses: We'll see the rise of more AI-native businesses designed from the ground up to solve specific user problems. Examples like #lovable, #cursor, #notebookllm, and #agentai, along with tight integrations with tools such as Gmail/Chat, will become more prevalent.
Vertical AI Solutions: Vertical AI solutions tailored to specific industries and domains will become more prominent, addressing unique challenges.
Reasoning-Focused Models: Smaller, more specialized AI models focused on specific reasoning tasks will be integrated into application solutions.
Integral AI Agents: AI agents will become increasingly essential components of software applications.
Alternative Interfaces: We'll see a move beyond chat-based interfaces, with AI increasingly operating seamlessly in the background or through other interaction modalities.
In conclusion, the fundamentals of business principles, customer focus, and product development will continue to drive further AI adoption, and we can expect that AI agents, AI-native solutions, and tight integration with existing solutions will be key drivers.
Any other trends are you seeing?