Dataiku has released a new version of its platform updated with GPT models with the aim of making it easier for enterprises to harness the power of OpenAI’s generative AI models while following responsible AI guidance.
The data science and AI vendor has introduced Dataiku 12, the latest version of its platform it calls “Everyday AI”. The platform is used to develop accessible AI applications, enabling users of various skill levels to prepare data for training AI projects without code.
Introduced on May 31st, Dataiku 12 comes almost a year after Dataiku 11, which featured a more focused integration for computer vision.
The Dataiku 12 update includes new features such as OpenAI GPT-4 integration. Automatic feature generation that is transparent and easy to understand. Model risk project view. The importance of universal function. and causal machine learning.
OpenAI integration allows organizations to incorporate GPT-4 models into their data projects using a visual interface and natural language prompts. Dataiku is not the only data science platform vendor working with OpenAI. DataRobot revealed in March that he was working with OpenAI to bring Azure OpenAI into DataRobot.
Universal Feature Importance provides a way to explain models to business teams. Causal Machine Learning helps organizations understand the reasons behind the results of AI models. The Model Risk Project View helps companies identify and mitigate risks in their AI projects. Finally, transparent automatic feature generation gives users transparency and control over the AI modeling process (the process by which AI models are created, trained, and deployed).
Mike Leone, an analyst in TechTarget’s Enterprise Strategy Group, said building in features into Dataiku that give users more ways to understand the process of an AI project will help move these efforts forward faster. Said it would help.
“Dataiku is focused on ensuring AI success by accelerating time to value,” said Leone. He added that this means vendors are providing safety guardrails to keep business users comfortable during AI projects.
Responsible AI
The new release of Dataiku comes as many companies seek ways to responsibly use generative AI tools such as OpenAI’s ChatGPT product. Generative AI tools are known to cause “hallucinations” such as spewing false information, and companies are wary of using tools integrated with generative AI.
“There is only general skepticism about AI, and generative AI.” Don Flackinger, Analyst, Enterprise Strategy Group Last month, in an interview about another generative AI product, he said: “What will that do for our business? Can I trust you? Some of these customers are probably mostly like, ‘Let’s see what we can do.’ . ”
It is this hesitation that Dataiku wants to address. But responsible AI is a difficult subject to tackle, says Leone.
“The biggest obstacle[to responsible AI]today is the lack of standardization,” he said.[toresponsibleAI}todayisalackofstandardization”hesaid[toresponsibleAI}todayisalackofstandardization” hesaid
Many organizations need help determining where to start on their path to responsible AI, he added. Organizations are also looking to narrow their responsible AI focus areas to explainability, bias detection, or fairness.
Dataiku has a multi-pronged approach with a focus on transparency, centralization and standardization, continued Leone. That’s why we give people involved in AI processes the tools to responsibly deploy and manage their models.
“They are focused on empowering customers to move forward with AI initiatives with confidence while ensuring trust and mitigating risk,” says Leone.
Dataiku isn’t the only data science vendor talking about responsible AI. On June 1, the Domino Data Lab announced new capabilities to help enterprises reduce model risk and use underlying models while implementing responsible AI practices.
Esther Ajao is a news writer covering artificial intelligence software and systems.
