Smile news

Walkthrough the State of the AI Report 2023

  • Date de l’événement Jan. 18 2024
  • Temps de lecture min.

The "State of AI 2023" report, published this week by Retool, offers a comprehensive overview of the current landscape of artificial intelligence in the workplace. Retool, known for its modern approach to building internal tools, is particularly relevant in discussing AI due to its focus on empowering developers and companies with efficient, cutting-edge technology solutions. This report, with its extensive data and insights, reflects Retool's position at the forefront of technological innovation and its deep understanding of AI's role in the modern workplace. Let’s dive into it and explore the metrics & trends.

Key Metrics

1. AI Usage Among Employees

A significant 88.4% of employees are using AI at work, intriguingly, half of these without the knowledge or support of their company.

This high percentage indicates a significant, perhaps somewhat organic, penetration of AI into the workplace. The fact that half of this usage occurs without formal company knowledge or support suggests that employees are independently seeking out AI tools to enhance productivity or solve problems. This trend might reflect a gap in company-provided resources or training in AI, leading employees to self-adopt these technologies.

Interestingly, respondents at those industry-leading companies were also the most likely to describe it as not useful, or sometimes “a total waste of time.” Opinions seemed to trend stronger the more familiar folks got with the technology—fair enough.


2. Internal AI Use Cases

Companies report that 66.2% have at least one internal AI application live, with popular use cases being code writing (47.5%), knowledge base Q&A (36.2%), and copywriting (32.9%).

The majority of companies are actively integrating AI into their internal operations. The presence of live internal use cases indicates a shift from experimental or theoretical AI applications to practical, real-world implementations. This could be driven by the need for efficiency, better data management, and the automation of routine tasks, all of which contribute to enhanced productivity and decision-making within organizations.

Contact us now for a free consultation and start discovering the benefits of experimenting with AI in-house. Don't miss the opportunity to transform your business with this revolutionary technology. Contact us today!


3. External AI Applications

About 43.1% of companies have at least one external AI application live, with in-product features (30.6%) and customer service chatbots (26.5%) being common.

The lower percentage of external use cases compared to internal ones may indicate a more cautious approach towards customer-facing AI applications. Companies might be prioritizing internal efficiency and process optimization before venturing into external applications, which often require more rigorous testing and quality assurance due to direct customer impact. This trend highlights a balanced approach in AI adoption, ensuring internal stability before extending its benefits to customers.

Ready to take your company's AI adoption to the next level? Contact us today to start implementing customer-facing AI applications and unlock a whole new level of efficiency and customer satisfaction. Let us help you extend the benefits of AI to your valuable clients. Don't wait, reach out now!


4. AI Hosting Preferences

A majority, 68.7%, use AI through hosted APIs, but half are considering a shift to self-hosted models. Larger companies, particularly those with over 5,000 employees, are more inclined towards this transition.

The preference for hosted AI APIs can be attributed to their ease of use, scalability, and lower upfront costs. However, the growing interest in transitioning to self-hosted models among half of these users likely stems from concerns over data security, control, and possibly cost-effectiveness in the long run. Larger companies, especially, might be more inclined towards self-hosting to ensure compliance with data privacy regulations and to have greater customization capabilities.

Ready to choose the best hosting model for your AI model? Contact us now and let our experts guide you through the process.


5. Future Plans for Open-Source AI

An emerging trend is the inclination towards open-source models, with 19.6% of companies planning to use them in the future.

The interest in open-source AI models indicates a shift towards more transparency, customization, and community-driven development in AI. Open-source models can provide companies with more control over their AI tools and potentially lower costs. This trend also suggests a growing confidence in the reliability and sophistication of open-source AI solutions, as well as a desire for collaborative innovation and less dependence on proprietary technologies.

Interested in exploring open-source AI models for your company's model making or finetuning needs? Contact us today to learn more about how our solutions can help you achieve transparency, customization, and community-driven development in AI. Let's work together to enhance your AI tools while potentially lowering costs. Don't miss out on the benefits of open-source AI – reach out to us now!

Additional Insights

  • Model Usage: Without any surprises, GPT-4 (38.2%) and GPT-3.5 (31.6%) from OpenAI are the most frequently used models, probably due to their performances and accessibility to the mass, in the convenience of an API.
  • Tooling for AI Applications: Hugging Face (25.8%) and LangChain (18.4%) are leading tools for AI app development. While Hugging Face is a clear key player, well known and here from a long time, the big adoption of LangChain is really encouraging for teams of developers and startups that would offer tooling solutions in this field.
  • AI Tools Used by Employees: ChatGPT (69.9%) and GitHub Copilot (68%) are widely used by employees in their work. The interesting part is that Github Copilot is nearly as used at ChatGPT. Again, that must be a side effect of the Retool survey base, but the Github Copilot adoption could be high due to the presence of a lot of developers in the panel.
  • Pain Points in AI Development: The main perceived challenges, for the companies, include model output accuracy (39.4%) and data security (33.4%).


The insights were gathered from a survey of 1,578 people in August 2023, primarily from the technology sector (39%) and spanning various company sizes, with 60% of respondents from companies having 1-99 employees. Due to that aspect, clearly influenced by the demographic of Retool clients, we should observe those numbers with a vigilant eye. However, we think it offers us a good glimpse of where the industry is going right now.


The report illustrates the deep integration of AI in various aspects of work, both internally and customer-facing. It highlights the growing trend towards self-hosted AI models, especially in larger organizations, and a keen interest in open-source models which is good news, aligned to the core values of Smile. The data also reflects on the prevalent use of AI tools like ChatGPT and GitHub Copilot among employees, indicating a widespread adoption across different sectors and roles of those 2 key players who managed to get a solid head start on the rest of the industry. This comprehensive analysis by Retool underscores the dynamic and evolving nature of AI in the workplace.

Thibault Milan

Thibault Milan

Directeur de l'Innovation