London Daily News

ChatGPT and competitive intelligence research

ChatGPT and competitive intelligence research

One of the common themes of the recent James Bond films franchise is that of an ageing James Bond is up against adversaries with a backdrop that the old ways are finished, and technology will replace him and his ilk. This article surrounds ChatGPT and competitive intelligence research. Tech is a wonderful thing and will transform our lives and sometimes for the better. This got us thinking about ChatGPT and the like.

In the age of increasing volumes of data, there are efforts to harness AI (or machine learning as it should be called) to make information collection and analysis “automatic”. OpenAI has created ChatGPT to do just this. Many questions have arisen regarding the role ChatGPT could play in secondary online intelligence collection and investigations. We have had a go at working with ChatGPT and other associated generative AI. Currently, there are many limitations to this technology. Oh, and we are not Luddites either and would like nothing more to remove the bits of the job.

Language

However, ChatGPT is restricted by the initial language input. The results will only be as good as what you ask it to do. And within competitive intelligence, the questions to answer are the most important aspect of achieving a successful project. Questions developed by people who are not engrained in your business or industry. Can ask the daft questions and come at a problem from another angle to define the real problems.

The information it can find and gather from online sources. There is little control. Like when you Google something, the quality and accuracy can be questionable. And sometimes, paywalls and deadends get in the way of research. And then ask other questions and look at things differently.

Bias

ChatGpt suffers from significant author bias. At best unintentional biases and, at worse ulterior motives and ever-increasingly sophisticated ways to create disinformation and deception. ChatGPT is limited by being unable to corroborate and evaluate information to the required standard with the increasing ability of deep fake technology. Culture, politics, experience, job roles and things get in the way of bias. A Silicon Valley-based Google engineer will think differently from a Texan oil worker, Peruvian school teacher or Indonesian hat maker.

Critical evaluation of the source’s quality, biases and context is something ChatGPT can’t do, and it may do in the future. Think about the times you just change the words you put in a search engine. Or the boolean strings available to focus results. You really have to use specific and concise language to get anywhere, and a one-word change can radically change the results.

One primary concern for researchers and, arguably more so, investigators surrounds OPSEC (operational security). Most projects are confidential, and every criminal investigation certainly is. Now, ChatGPT uses Machine Learning. Meaning the platform is learning from what you are doing and asking for. Does this mean your work will find its way into the open-source world? It may be if someone else asks a similar question. Further research needed on this aspect of ChatGPT is required.

A good analyst and researcher will be asked to answer a question. Great, but 90% of the time, the analyst needs to be creative in their analysis and be about to think laterally.

Analysis

ChatGPT is currently unable to critically evaluate the quality of the source like an experienced analyst or investigator. ChatGPT currently struggles with lateral thinking and going beyond the information. To connect the dots, as some insist on saying. To continue the dot analogy, if ChatGPT and the like can’t connect the dots, it will have even less success determining the spaces between the dots and what’s not being said or reported.  

ChatGPT is limited beyond information gathering and oriented toward the initial commands and specific data points. Currently, it can’t do network analysis, character profiling of a person, and pattern detection. Analysts need to be knee-deep in their information to get a picture. ChatGPT and the like may eventually deliver an excellent end product, but will the report’s author really understand what’s going on? The nuance of what’s happening or not happening.  And this thinking leads us to ask if you would be happy for a ChatGpt-type platform to decide whether to pull the trigger or not. OR better still, to push the nuclear red button or not?

Language-based AI technology struggles to isolate and see the nuances of human emotion, especially baselining and monitoring shifts in behaviour and sentiment.

ChatGPT and competitive intelligence research

Finally, an aspect of investigative work is the accurate recording of an investigator’s workflow and cognitive processes. In law, investigators must present their cases in an ordered and acceptable way. They need to prove their workings and sources if a judge or barrister asks them to do so. “I don’t know I just pressed this button and this was the result” will not be the answer they will seek.

The end product is currently soulless and passionless and will struggle to replace quality researchers, analysts and reporters. But it will save them a lot of time. It may well turn out to be an irreplaceable tactical tool to assist in data collection. Still, it will not replace the analyst and investigator of synthesising and analysing the information and creating intelligence.

We are Octopus, and we’re all about your real competitive advantage.

We are the global bespoke people-powered global competitive intelligence company creating the insight you can do something with.

Follow Us

Pin It on Pinterest