The March of AI

Generative AI (artificial intelligence) is the latest step in the evolution of complex, high performing algorithms that can generate content. While generative techniques have been around for a long time (AI is not a new thing), with the advent of large and powerful artificial neural nets (ANNs) they have exploded into the public imagination with chat bots such as ChatGPT, closely followed by open source models such as Facebook’s Llama.

What does this mean to the field of product, project and programme delivery?

Generative AI

Think back to the arrival of the internet: search was a thing, and a hotly-contested battleground. Over time, search matured, and there was even a time when search engines were remarkably good - able to direct you to relevant websites with high quality content. However, as the internet has been monetised, search engines have - in general - re-prioritised money-making results, including adverts and biased content, in which the user sits in a content filter bubble.

Now, though, with generative AI that’s powered by large language models (LLMs), we’re seeing another change. The generative models provide a new type of ‘search’, where the large language model (LLM) is trained on very large datasets (essentially, subsets of the internet itself, not always respecting copyright either). The LLM can then engage with the user in a natural language interaction, and answer queries, synthesising answers that are, in a sense, the search result. They have become not only the search engine, but the search result, morphed into one.

Large Language Models are not the only type of generative AI. There are image-based techniques, such as Stable Diffusion, that are able to extract the essence of an image style, and re-apply it to a new image, or even generate images from scratch based on a prompt, combining language and image analysis techniques.

Delivering with Generative AI

Projects, programmes and portfolios (whether agile, iterative, waterfall, hybrid etc.) are all benefiting from generative AI, at a number of different levels.

For example:

• Lessons learned: powered by LLMs trained (fine-tuned, in the parlance) on specific project documentation
• Summarising meeting minutes, and identifying actions
• Automatically deducing ‘ways of working’ documentation - building custom quality management systems
• Predicting outcomes through predictive analytics (such as Project Science’s Foundations suite)
• Generating marketing and communications content

However, there are challenges, including:

• How do we ensure the generative techniques do not leak data into the public domain?
• Avoiding bias in the results
• Quality control of the outputs - LLMs can give the wrong answers with great confidence
• Making sure that people stay in charge and in control - we cannot have “computer says no”, without at least knowing why
• Legal, ethical and regulatory compliance

Gaining help

We can help your organisation understand what Generative AI means for delivery, and how you can implement it. There’s potential in every part of your business, including:

• Strategy
• Marketing
• Sales and lead generation
• Planning and Operations
• Portfolios
• Programmes
• Projects
• Customer Support
• Continuous Improvement

Get in touch with Project Science to find out how we can help your business realise the enormous benefits of generative AI.