Q: ChatGPT – Angelic or evil?
A: Neither. If you haven’t experimented with ChatGPT, it is what computers do in the movies. Ask a question, get an answer.}
ChatGPT is a tool. It is not inherently endowed with good or evil.
When initially experimenting with ChatGPT, I was reminded of the first time I saw a browser or saw Java. The possibilities are clear—the initial implementation is limited. If you don’t believe me, ask, “Who is the current president of the U.S.?” Now, I understand the limitation in answering these questions is due to the training of the corpus, but the why isn’t very interesting.
The fact is there are real limitations. Ask what day ChatGPT stopped learning and again get an unsatisfying answer. Conversely, ask for an essay on the Canterbury Tales, and prepare to be amazed.
DALL-E is similar. It’s scary fast, but studying the output will reveal real limitations. A little tiny of bit of domain knowledge will suffice to realize something is amiss. If you have a sense of what Bruce Springsteen looks like, you won’t be fooled by anything that DALL-E conjures up involving the Boss.
There are three realities to consider here: (1) Tools such as ChatGPT will improve; (2) Today, many CEOs are asking their leadership teams, “What is this? What does this mean?;” and (3) These tools are not going away. The genie doesn’t go back in the bottle.
So ChatGPT is neither good nor evil—it is a tool with both positive and negative aspects
Negative Aspects: Bad people getting assistance doing bad things.
According to CyberArk, the negative aspects have started. But let’s add a caveat to CyberArk here: There is commentary indicating the referenced blog was produced by AI, although this may be tongue in cheek. In the blog, it is noted that ChatGPT’s ability to produce code has malicious actors producing polymorphic malware that can avoid signature-based detection defenses. None of this is new. There is a reason that other defenses exist.
Yet, this remains a real concern. If relatively unsophisticated people can get assistance generating risky code, how much damage could a very sophisticated person do with ChatGPT-type tools? Paradoxically, but not surprisingly, the only solution to this particular problem is increasing the utilization of AI-driven technologies in defenses.
The code produced might be crude and limited, but malicious actors don’t care about passing an enterprise review board or if more than 99% of their code is a total failure. One success equals ultimate success.
Increasing the productivity of malicious actors is just one example of a negative outcome; deepfakes and corpus poisoning are two others that come to mind. All three of these examples show why I confidently state the genie is not going back into the bottle.
They also exhibit why CEOs are correct to ask their organizations, “What does this mean?” No organization is going to write a policy to ban utilization and be immune. ChatGPT and similar tools, even their negative aspects, are a reality all parties are going to have to face.
Positive Aspects: Let machines do mechanical things/let humans do human things.
It’s not all bad news, though! Some aspects of the rise of ChatGPT will be positive. Many people have noted how quickly essays can be written, how quickly basic code can be produced, and similar activities with alarm. My view on these elements? When used appropriately, great! Do these activities really add value in the 21st century?
Looking at my own role, if ChatGPT can produce basic briefs or blogs better than I can, then so be it. My value, even more so than today, will come from providing insights born from years of experience tempered with current marketplace forces, the ability to network and to share networks, and the ability to be thought-provoking.
In essence, my value will come from being more human.
This will put increasing pressure on knowledge workers. Anything that could be produced, even if it wasn’t, by an AI tool will be commoditized. The upside, however, is that the things ChatGPT can’t do—the human talents of engaging, networking, and being uniquely insightful—will become more valuable than ever.
So to answer the original question: Is ChatGPT inherently good or evil? It’s neither. It is a powerful tool with both negative and positive aspects like all technology before it.
Have a different opinion on ChatGPT? Let me know at [email protected].