TIME TO READ: 3 MINUTES

Open the magic box of Generative AI for people to trust the technology

Continuing my AI journey, I recently completed a very enlightening course from IBM on Generative AI: Impact, Consideration, and Ethical Issues.

One of the main themes was the importance of transparency around the new, and rapidly developing advances that seem to appear every day in the Generative AI technology space. There has been a rush to claim to be first to the table with this exciting technology. It can seem overwhelming and frankly scary to people who don’t fully understand.

By opening the “Magic Box” and explaining in non-technical terms, the benefits, risks and limits, people will have more confidence that this is a technology to be embraced and not feared.

The responsibility for being more transparent lies first with the creators of the technologies but also with the businesses using generative AI. There is a need to be open about how people’s data is being used and for what purpose. People need to feel reassured that their personal data is secure and that copyright and privacy laws are not being breached.

In business, it is important to have a governance policy on the use of Generative AI that covers these issues. Currently, there is a lack of AI regulation but that could and mostly will have to change in the future, so it pays to put guidance in place. Identity fraud, misinformation, copyright infringement and data privacy violations are all issues that need to be considered and a good governance policy will migitate.

Biases can occur when the data generative models are trained on have not been checked and vetted. Ideally, data should be “clean” without discrimination and biases. However, the majority of LLMs (large language models) have been trained on the content of the internet which we all know to contain the best and the worst of human ideas.

The constant checking of outputs by humans should be the norm, with methods in place so users can flag issues.

Trust will only be achieved if the “Magic Box” of Generative AI models are opened up, as reliable and transparent tools for their users.

---

What responsible use of Generative AI is centred around:

🔍 Transparency
Openness as to how it all works. People need non-technical explanations to trust the “Magic Box” and understand its potential benefits, limits, and risks.

⚖ Accountability
Holding individuals and organisations responsible for ethical and legal consequences relating to copyright and privacy laws.

⛑ Safety guardrails
Constantly checking for biases in data, and being aware that hallucinations can occur. Building in error handling where users can report issues and action can be taken.

⛔ Privacy
Taking steps to protect personal data, complying with data protection laws and ensuring generated content does not disclose confidential information.

👩‍🎓 Training
Upskilling your workforce. Jobs are going to change and new jobs will be created. By investing in your staff, you give them the best opportunity to embrace this technology and increase productivity.

#Ethicalai

©Jennifer Martin 2024