ChatGPT: be careful what you say?
Photo by Mojahid Mottakin

ChatGPT: be careful what you say?

Heard of ChatGPT? I get asked this several times a day.?

Some of our followers have?been asking us whether it presents a threat to what we do at SimSage. In actual fact, it's just created what looks to be a significant opportunity for us.?

There's no doubt that ChatGPT is clever, but at what? For the moment, we consider it to be well up on the 'Hype Curve'. Everyone is rushing to try it out and see how cleverly it constructs text, prose, poetry etc. Apparently, it even writes code. Hmmm...?

Things that are worth remembering though:?

  • ChatGPT assembles 'factoids', pieces of data/information in a way that 'looks' very well-written but may (often) actually be factually wrong. It doesn't 'know' if what it writes is correct, so that requires the person using it to know their subject matter very well. Imagine a lawyer producing a document that is actually incorrect. “Impossible”, you say? Well, apparently it's already happened. That's not the 'fault' of ChatGPT. It's done what was asked of it. It just doesn't 'know' if what it has written is right or wrong.?
  • It works 'statistically', in that, while creating a 'piece', it takes the words that 'most likely' follow-on from the ones it has already been fed or created. Uh oh! Some of the results people are posting online are very funny; others are simply disturbing.?
  • ChatGPT is highly prone to being fed mis/disinformation. It will spew out what it is fed, which might work well for those whose intention it is to promulgate that sort of output.?
  • Currently, it doesn't reference where it finds the information, so good luck with that.?

Will they be able to fix all the above? Well, they are backing themselves to do so, and so, it seems, are companies and people like Microsoft and Elon Musk. There's clearly some way to go right now.??

Notwithstanding all that, it's fun to play with, but be very careful what you feed into it! There's been an outcry this week from the Italians who are concerned that ChatGPT may be infringing on PII, and also IP (Intellectual Property). They have banned use of it in Italy for now and are reportedly taking action against ChatGPT (see more).?

The EU is working on legislation to control the use of Large Language Models (LLMs) like ChatGPT on account of similar concerns about how it actually produces its output.?

The way I see it, AI can be incredibly useful, but also damaging. We use AI 'components' at SimSage, simply to speed-up or automate processes. SimSage doesn't make things up. Everything we process is reference-able back to the source document and information. Plus, we specifically find and reference PII, including names, addresses, email addresses, phone numbers, bank/credit card information, NHI numbers and so much more.?

So, nope - we're not at all threatened by the ChatGPT technology. In fact, they have created a whole new world of opportunity for SimSage.?

If you're concerned about whether you are GDPR/PII compliant, or you think you've suffered a data breach, we're here to help. Find out more about our technology on our website (www.simsage.ai) or contact me directly.

Stephen Wray

Senior Security Consultant

1 年

Its's like nuclear technology; it has opportunities and also threats... Businesses really need to start putting policies in place around it, to start with. Eg; Do not ever put anything in a Chat GPT prompt that should not be public.

要查看或添加评论,请登录

Sean Wilson的更多文章

社区洞察

其他会员也浏览了