Amazon’s CEO, Andy Jassy, has identified generative AI as the company’s next big focus. In his annual letter to shareholders, Jassy outlined Amazon’s plans to position itself at the forefront of the AI revolution. The company is highlighting the three layers of its GenAI stack. The bottom layer focuses on training models and reducing customer costs, the middle layer allows companies to customize existing models, and the top layer involves building generative AI applications for Amazon’s consumer businesses.
OpenAI has released an updated version of ChatGPT, GPT-4 Turbo, which offers improvements in writing, math, logical reasoning, and coding. The new model uses conversational language to provide more direct and less verbose responses. It also has an updated knowledge base and was trained on data up to December 2023.
Also, OpenAI’s GPT-4 Turbo has reclaimed the “best AI model” title from Anthropic’s Claude 3 in the Large Model Systems Organization (LMSYS) Chatbot Arena. The updated version of GPT-4 Turbo, gpt-4-turbo-2024-04-09, is leading in the overall coding and English categories, showcasing improvements in coding, math, logical reasoning, and writing capabilities.
According to a report by Productiv, ChatGPT, an AI-native app developed by OpenAI, has become the most prevalent shadow IT application, surpassing other popular tools like LinkedIn, Canva, and Adobe Acrobat.
From The Decoder, OpenAI CEO Sam Altman is pitching ChatGPT Enterprise to large corporations, highlighting the advantages of working directly with OpenAI, accessing the latest models, and having more options for custom AI products.
According to Gartner research, by 2028, three-quarters of enterprise software engineers will use some form of generative AI code assistant. While these tools can lead to increased job satisfaction and retention, the actual efficiency gains may vary. Organizations may only see 5-15% efficiency savings when using coding assistants on complex business logic and legacy codebases. Additionally, AI tools cannot challenge developers like human pair programmers can.
Why do we care?
Amazon’s approach offers potential opportunities for providers, too—training models, customizing existing models, or building applications. I’d be remiss if I didn’t include implementing applications, helping with policy and data management, and ongoing system monitoring. OpenAI sees that, too—they have an enterprise sales strategy, and I wanted to highlight that.
The popularity of ChatGPT as a shadow IT application indicates a grassroots-level acceptance and utility perception among end users, which contrasts with formal adoption strategies at the organizational level. This suggests a need for companies to align their IT strategies more closely with actual user preferences and practices to harness the full potential of AI tools.