We help our clients avoid ethical, reputational, and compliance risks using our bespoke auditing and ethical risk and impact assessment framework.
We consist of professional academics that contribute to cutting-edge research in responsible innovation and machine intelligence.
We use proven strategies cultivated through years of experience in the space of organizational ethics consulting and change management.
How we can help you
Responsible AI Governance Gap Analysis
This is a targeted assessment of the client's current Responsible AI practices and governance in relation to current (and emerging) standards, regulations, and best practices.
The practices assessed include AI development processes, documentation, reviews, testing, monitoring, stakeholder engagement, and external communication and transparency.
Algorithm Risk & Impact Assessment
This is a deep assessment of a particular algorithmic sociotechnical system to identify ethical, compliance, safety, liability, and reputational risk. A full report and recommendations for mitigating those risks are provided.
Algorithmic Bias Assessments and Audits
This is an assessment of potential bias in an algorithmic system for internal development purposes, though results can be shared publicly if the client wishes. To accomplish this, we makes use of inputs from our algorithm risk and impact assessment and highly technical tools and expertise, including:
- Proprietary algorithms for evaluating bias;
- Carefully curated test datasets for each industry/use-case;
- A world-class team of experts that are helping define this burgeoning industry
Bias audits are a stringent independent review of potential bias in an algorithmic system that is shared publicly and based on 3rd party frameworks.
Corporate Training in Responsible AI
This involves developing and delivering training and compliance courses in Responsible AI to fit corporate upskilling needs.
Example areas may include conducting technical bias audits or ethical risk analyses based on proprietaty BABL methodologies.
Case Study: Leading AI Vendor
Public and regulatory pressure
A leading AI vendor comes under intense public and regulatory scrutiny for potential bias in its core AI product, eroding the trust of clients.
A Responsible AI Strategy
Through a multi-workstream project, BABL AI helps the client:
- Conduct a bias assessment of their core machine-learning algorithm;
- Develop and execute an Ethical Impact Assessment, identifying key ethical risks and governance mechanisms for mitigating those risks;
- Develop data collection and labeling best practices, with a focus on mitigating potential sources of societal bias in training data;
- Implement a data quality and model monitoring program, with demonstrated success measured through continuous improvement and reduced bias in their production algorithms.
Regain of trust with public, regulator, and key clients
Through iterative improvement and transparent bias assessment documentation, the vendor is able to build public trust, address regulatory inquiries with good-faith, and retain critical clients.