Procurement’s AI Frontier: The early adopter gets the digital worm

Posted: 31st August 2023 Spencer Hill, Director of Commercial Services

In his recent article ‘Why we all need to start thinking about Artificial Intelligence now’, Michael Appleby presents some startling facts and figures about the growth of this disruptor, and cautiously estimates that:

“Right now, every organisation in the country has at least a small group of employees who are actively using some form of AI tool to support their work – mostly unbeknown to their managers.”

He then shares his view on the limits of AI and provides some thoughts on what organisations need to be considering right now. Michael proposes that AI:

  • Is not yet perfect (but it is learning fast).
  • Represents potential data and governance breaches.
  • Is growing from the bottom up and is uncontrolled.

This article continues our series on AI and takes a commercial and procurement view.

Within the world of Procurement, AI is not a new topic, and I remember running a workshop in 2017 which we based upon KPMG’s Future of Procurement report published the previous year. The report predicted four futures for the procurement function plotted onto axis of humans to algorithms and centralised to decentralised. When algorithms rule in a decentralised world, the report predicted a future they entitled ‘RIP Procurement’, which is quite a chilling title.

We asked the room full of procurement professionals to imagine they were in the future and asked them to discuss how they would adapt to this new way of working where much of their role was automated, and computers and clever algorithms were advanced and widespread.  Mixed in with the sensible suggestions we gathered that day was fear mixed with a considerable amount of nervous laughter. This was scary, but it wasn’t going to happen during our lifetimes, right?

Only six years later, it is clear we were all very wrong.

So how is AI going to impact on our world?

In brief:

  • It can generate content for you.
  • You will likely receive generated content.
  • It could increase your productivity.
  • It potentially increases your risk of data breaches and challenges (depending on how you use it).

Content generation. The exciting news is that once you get the hang of using prompts (the instructional language you use to communicate with the AI interface) you can generate anything you can think of. Everything from procurement strategies to contracts to specifications to tender packs to responses to tenders and much more. The only limits are your imagination, subject matter knowledge, and experience with the interface. You can set the word limit, link to existing documents for review and summary, and even describe a role for it so it knows how to phrase the answer. Be cautious though. I have yet to strike content gold on the first attempt, and even after several iterations and reworded prompts, I will still end up with a document that is a starter for 10 rather than the finished output. Also, knowing I have a starter for 10 relies on my experience to spot this. I could select a random person off the street, teach them the prompts and they could generate a beautiful procurement strategy, but they would have no clue if it could be used.

My advice? Use it for research, to break the blank page curse and possibly to summarise something which is TLDR (internet speak for Too Long Didn’t Read), but whilst it is in its infancy, scrutinise everything it produces before distribution.

You will receive generated content. This is probably already happening, and it is unlikely you will spot it unless the author makes the mistake of taking the first draft and sending it without proofreading. You will already be receiving CVs, tender responses, emails and much more which has been generated by AI. Some producers are proudly announcing that they wrote their content themselves, but how would you know?

My advice? Don’t worry about it. Continue to review everything you read to ensure it is fit for purpose and deal with errors and blatant plagiarism as you always did.

Increase your productivity. Studies are already suggesting that productivity can increase as high as 40% using AI tools when you use it to produce content. It is very quick and as I have already described, there is nothing it won’t have a go at, so why not ask it to write you a 500-word summary of the ground source heat pump market in the UK, or an RFQ to purchase a Landscape Architect for a public sector organisation. They will both need lots of work before you could use them, but AI has at least got you started. You can also use AI for advice, to solve problems to proofread or to make decisions, and I don’t doubt that you could feed it a question then a series of answers and ask for an opinion. You see where I am going.

My advice? Read the data breach part first, but why not use AI to do your research for you, generate some filler content, lay out a structure for a report? There is certainly no harm in using the TLDR prompt to summarise something which you haven’t the time to read yourself.

Here are a selection of prompts that a friend has sent to me which were originally tweeted by @shedntcare_ claiming they will make you 10 times more productive than Elon Musk. If you are new to AI and looking for examples of what it can do, try these:

  • Tell me the 10 important questions for beginners about (insert topic). Create a table with examples of the questions.
  • Your job is to solve problems. Give me a step-by-step guide to solve this problem (insert)
  • Summarise the book (insert) by (insert author) and give me a list of the 10 most important teachings and ideas.
  • Proofread my text below. Correct any grammatical or spelling errors. Make suggestions to improve my writing. (Insert your text).
  • Summarise the following text in 500 words or less. Create sections for each important point with a brief summary. (Insert text).
  • Act as (insert famous character). Give me your opinion critically. My text: (insert text).

The risks of data and confidentiality breaches. Quite simply, everything you load into these tools is uploaded somewhere to enable it to be processed, so be mindful of what you share. As procurement professionals we would not release potentially sensitive information relating to bidders within a public sector tender to the public domain, so the same rules apply here. The same goes for anything you would not want to be leaked.

My advice? Use your common sense whenever you use these tools and exercise caution. Don’t upload something you would not otherwise share on public forums, and that includes text and naming individuals or companies in any way. Undoubtedly there will be some specific procurement related policy from Government soon enough, and there is already Central Government guidance produced in September 2022 through a National AI Strategy which is worth a read, particularly Pillar 3 Governing AI effectively.

AI is here, and as everyone who writes about it will suggest, it is probably already in use in your organisation or by your stakeholders. Trying to control it with draconian measures will only drive it underground, so why not be open about it and ensure that it is being used intelligently with organisational guidance and training.

As AI grows it will become more reliable, the limitation on its knowledge will lift, and the quality of the outputs will match its own confidence in them. I remember that our conclusion in 2017 was that regardless of how clever the machines got, they would still need humans to operate them. I still believe that to be true, so our objective is to become the best operators we can and ensure we are in control and not the other way around.

If you want to learn more about AI and how it might impact the industry, or if you’re interested in discussing how to bring AI into your company’s commercial and procurement strategy, please get in touch with:

Spencer Hill

Director of Commercial Services

Latest News

See all news