Nebuli’s Augmented Intelligence Quotient project (AIQ -pronounced “IQ”) offers cleansed and trustworthy large language models to help businesses and institutions deploy responsible AI systems faster with significantly reduced biases, misinformation and harmful content.
AIQ is our response to the increasing use of AI-based services, which pose dangerous wider economic, social, environmental, cultural, and political outcomes through misinformation or bogus human-like chatbots. Not to mention the increasing concerns over data privacy, security and the potential negative impact of AI biases on marginalised communities.
This presents the opportunity for AIQ to establish a new standard that ensures AI systems are fair, transparent, and aligned with ethical principles. We aim to apply LLM models integrated with the upcoming AI regulatory frameworks, such as the EU’s forthcoming legal framework on AI.
By combining AIQ’s data models with the key emerging technologies, teams and enterprises can focus on augmenting human decision-making and creativity, rather than replacing it.
From our experience in the market, the following are the critical challenges that impede teams, SMEs and corporations’ successful development and deployment of responsible (“ethical”) AI/ML solutions:
Working with inaccurate or incomplete datasets can lead to poor model performance and unreliable results.
Due to the current severe data science workforce shortages in this domain.
Lack of standards covering ethical and legal issues relating to bias, privacy, and accountability.
Many AI and ML models operate as “black boxes,” making it difficult to understand how they arrived at their decisions.
Handling large amounts of data and high performance requirements are challenging and expensive.
AI and ML models can be vulnerable to ML-based attacks.
With AIQ, we are exploring mitigation strategies for the above challenges, alongside current state-of-the-art approaches — such as explainability and interpretability methods, Fairness, Accountability, and Transparency (FAT) algorithms, Human-in-the-loop (HITL) techniques, and AI governance and oversight frameworks.
We have already addressed some of these needs by researching and developing solutions for our enterprise customers’ involving responsible AI, data ethics strategy, AI governance and oversight frameworks and with AIQ we aim to extend this opportunity to SMEs.
We constructed AIQ using Nebuli’s Datastack’s data and cloud integration framework, which allows teams to untrap their data from multiple sources combining them into complete and accurate datasets for digital business processes and sophisticated data analysis.
We help customers employ our Datastack framework to integrate the traditionally separate business-critical data services, such as data security, compression, modelling, classification, segmentation, knowledge discovery and much more, into an API-powered integrable service.
The AIQ project helps us expand this framework through a public data lake and use dedicated API gateways to specific datasets targeting specific tasks and sectors.
We are exploring three key “transactional” interaction models to take place through AIQ:
Developers and SMEs using an AIQ console to securely access and connect to their targeted datasets. They can also use the same console to contribute their open datasets and algorithms for crowd-sourced improvements.
Nebuli’s existing corporate customers with higher demands for more complex and multi-sector datasets would use Nebuli’s services to apply customised and dedicated client API gateways for their intern R&D projects.
The ability for machine-to-machine transactional interactions, where smart machines can access specific datasets and contribute datasets to the network autonomously. NOTE — We expect this feature to be highly experimental and will require a much higher level of scrutiny against bias and ethical risks. Thus, it is more of a vision than a plan for the short term.
AIQ project’s mission is to make AIQ readily available for individuals, teams and organisations of all sizes to assist them with their research and development projects, technology-based innovations, productivity challenges, augmented creativity and much more.
Our vision for AIQ is to evolve it over time into a decentralised network for repositories of large data models, LLMs, code libraries, design frameworks, tools, and algorithms that collectively contribute toward helping teams, SMEs and enterprises build human-centric and responsible augmented intelligence applications.