Google just launched the most powerful large language model (LLM) the world has ever seen – Gemini. Gemini now powers Google’s publicly accessible genAI service: Bard.

Last week the UK government published a report claiming that the finance, public, legal and education sectors would be the sectors most severely impacted by genAI.

So we asked Bard (Gemini) how those sectors will be impacted by AI, and we then asked our sector experts and our AI director, Luke Budka, for their opinions.

How will the public sector be impacted by AI?

Bard says:

GenAI has the potential to transform the public sector by enhancing efficiency, improving service delivery, and making better informed decisions.

It can be used to:

  • Analyse large datasets: GenAI can analyse vast amounts of public data, such as census data, social welfare records, and traffic patterns, to identify patterns, trends, and insights that can help policymakers develop effective public programs and policies.
  • Automate tasks: GenAI can automate routine tasks in various government departments, such as processing benefit applications, reviewing permits, and conducting investigations. This can free up public employees to focus on more complex and critical issues.
  • Enhance citizen engagement: GenAI can analyse social media data, online comments, and surveys to gain insights into public opinion and identify areas where citizen engagement can be improved. This can help governments tailor their services and communicate more effectively with their constituents.

Fiona Corey our MD of public sector, education and infrastructure adds:

“When it comes to enhancing citizen engagement, accessibility is a big part of this. Translation tools, audio transcriptions and so on, will all help governments reach a wider audience.”

What should the public sector do about it?

Luke Budka, our AI director says:

“While there are obvious benefits to large scale data analysis for the public sector, there are easier gains to be found in straightforward summarisation and analysis tasks.

“The UK government’s initial experiments with summarisation of local skills improvement plans (LSIPs) across 38 areas of England are a good example of this and showed promising results: the vast majority (75%) of individual LSIP summaries produced by the AI were either wholly or mostly correct. What was arguably more encouraging was the prompts used to extract and summarise the LSIP reports were successfully adjusted throughout the pilot based on feedback from the LSIP policy team. It is therefore easy to surmise that training civil servants in prompt engineering will clearly deliver dividends – as the report states: “…the success of generative AI is directly related to the quality of prompts provided.”

“But for government, it’s going to come down to affordability first and foremost, and that’s where the worry creeps in. For the eight-week LSIP summarisation trial, participants were equipped with access to the latest OpenAI model, GPT-4.

“In reality though, equipping significant volumes of civil servants with this tech will be cost prohibitive, unless done in a private environment where the government only pays for API calls, rather than individual user subs. This will become even more affordable as LLM model prices drop/open-source models demonstrate parable performance.”

Fiona adds:

“Aside from affordability, confidentiality is also a huge issue for the public sector and private environments – that keep data confidential and secure – would be a pre-requisite.”

 

Drop us a line if you’re interested in having chat about our genAI services, from tone of voice tuning to private environments.