Here’s a thought

Every day SkillGate tutors mark hundreds of assignments for our customers. We award certificates based on the quality of their written assignments.

At the moment, we are seeing around 10% of assignments flagged as including 80% or more AI generated answers.

In general, due to the style of SkillGate Certificate programmes and the way we structure our questions, our students still have to complete the course to be able to complete the assignments.

However, AI agents are getting through – they can take a course, pass the tests and write assignments (with perfect or even imperfect grammar if requested!).

This will become a problem.

Does it matter?

Let’s call it the Red Dwarf problem (after the popular TV series). Holly is the eponymous spaceship’s onboard computer. Holly solves all the real problems while the crew plays around messing things up.

Holly (AI) represents sensible and good, the crew (staff) are bad. The TV programme is a metaphor for how we view technology.

But it is not real. Red Dwarf has to be placed 3 million light years from home to make it believable. It is funny because it does not matter if the crew do stupid things or is even obliterated. They are not going anywhere. The regulators don’t care what they do. They have no customers.

Your regulators and customers do care what your staff do.

The rule of unintended consequences

It is the rule of unintended consequences where AI technology is already delivering tangible examples.

Let’s take recruitment. Market leading recruitment companies initially opened up the recruitment world by trying to simplify the matching of CVs to recruiters. It was so successful that now recruiters receive thousands of CVs for any job – good effort – except of course that it isn’t.

So now recruiters use AI to sort through the thousands of CVs. This is great except that the applicants now also use AI to write their CVs. We have machines analysing the work of other machines and we will soon be back where we started.

A successful City recruitment firm told me recently that recruitment for junior staff has more or less dried up. City firms are using AI to make recruitment decisions.

Are they aware of the effects of marginalising their own managers? They are letting computers build their next generation of teams. If they are doing this to save a few dollars – is this wise?

Stop AI doing our courses and taking our tests

Like everyone else, the training industry is fixated on using AI. There are a number of potential opportunities especially around the creation of individualised or adaptive learning solutions.

This definitely has value and SkillGate has offered a series of developments in this regard over the last few years and will continue to do so. But the arrival of AI means first we need to understand how we will stop AI ‘reading’ our courses and passing the tests for our learners.

Do we want our children’s teachers doing their Safeguarding training by tapping the questions into ChatGPT and copying the answers? Do we want our Care Home workers getting Copilot to do their Dementia Awareness training?

This will happen.

At the moment, it may be possible to know technologically if an AI agent has taken the course for you. But we have to assume this will not always be the case. Eventually ‘Holly’ will circumvent any controls we put in place.

They call it ‘artificial intelligence’. It may be intelligent but it has no imagination. A truly intelligent system would turn around to the teacher, or care worker, and say ‘No, that is not right. You should not cheat like this. I won’t help you’ – but it won’t.

So how do we, and the regulators, know that a member of staff really does have the knowledge and skills we have tested them for?

The rise of the trainer

There is only one way. Your human coach, assessor, judge, trainer or manager will have to make informed decisions based on their expertise and experience. They will need to judge, mark and record actions or answers (ironically – you may think – like they used to).

We can anticipate more formal coaching assessment (check out our CoachAssess software), more invigilated testing (see SkillGate’s Classroom) and more reliance on our line managers to check knowledge and skills (try SkillGate’s SkillsAudit, 360 Feedback or TLSignOff).

Our latest product, TLSignOff, for example, warns learners taking SkillGate tutorials, that their Team Leader will conduct a secondary check on their knowledge. On completion, the course will send a short list of verification questions to the line manager for checking with the learner.

Advanced learning, training and performance systems can use AI, but they will also put the emphasis back where it should always have been – on humans, as trainers, coaches and team leaders, to make and record assessments.

Teams matter, and (to corrupt a popular saying), there is no ‘AI’ in Team.