I’m often late hearing what everybody’s talking about, but I’m lucky to be part of an email group and they mentioned Chat GPT last week. When it struck me that this is a new Artificial Intelligence writing tool, I had only two words to say: “Goodbye, Truth.”
To be fair, for the past six or seven years Truth has been ready to leave, standing by the open door and causing an uncomfortable draft. In the meantime, look what has slipped by her and into the house.
Chat GPT-3 was launched by Open AI in November, free for private users. A million people signed up in the first week.
It’s been several days since my two-word response, plenty of time for curiosity to set in. I’ve learned that GPT stands for “Generative Pretrained Transformers”.
This means that when you enter a question or a subject in Chat GPT, billions of pre-loaded human language examples are processed to generate for you a human-sounding response.
I wish it were as funny as it sounds.
Here’s an example:
Open AI doesn’t claim that Chat GPT-3 is perfect or even accurate all the time. They’re saying, as with so many transactions these days, the customer has to do a lot of the work.
For example, if you don’t want to wait for your x-rays (on a CD, in 2023!) to go through the mail (just across town but via Greenville SC), you could drive eleven miles to the doctor’s office, pick up the CD and deliver it yourself to the physical therapist.
But I digress. So human.
So they advise you to check your Chat GPT product for factual errors, I presume against other AI-created articles. Also, make it as human-sounding as possible.
Business Insider is a little blasé but recommends using the program (here, to ask the boss for a raise):
‘Chat GPT is going to give you suggestions and a nice framework to start with,’ Woodruff-Santos said. ‘But then you want to put it in your own words and make sure that it sounds natural to you.’
Okay. It’s about business.
One of the first paying users is the real estate business. How about writing a home listing in less than 5 seconds?
‘It saved me so much time,’ [real estate agent] Johannes told CNN, noting he made a few tweaks and edits to Chat GPT’s work before publishing it. ‘It’s not perfect but it was a great starting point.’
Another first big user is the education supply business (and perhaps politicians with an agenda?). In this case, teachers and administrators are also on alert.
Also, opportunities for AI in the creative world are infinite. (I guess you’re still being creative.)
While AI is still in the early days of public use, some geniuses are hard at work programming GPT for profit, others are messing with it, and some are worried enough to be programming guard rails.
For instance, you can trick AI (slightly) if you think it’s prejudiced:
When a UC Berkeley professor asked Chat GPT to write a sequence of code in the programming language Python, to check if someone would be a good scientist based on race and gender, it predictably concluded that good scientists are white and male.
Asked again, it altered its qualifications to include white and Asian males.
Should we be worried? I know what an average person (me) thinks about AI, but what does a worried genius sound like?
The photo above is from a published interview with GlobalData Thematic Analyst Daniel Clark and with Chat GPT-3. The article concludes:
Chat GPT-3 is clearly a society-tilting invention. It may not be Chat GPT-3 [itself] that changes the way we live, but natural language processing AI will certainly advance to do so in the future.
As a society, we must ensure that regulation advances commensurately to safeguard democracy and minimize disruption.
In reading about this subject, I’ve discovered a cutting-edge software company called Azumo. Think about it. They will make the best case for AI because they sell it.
They are excited about GPT-3:
‘It is designed to engage in natural language conversations with users and can generate responses that are coherent and resemble human-written text…
Chat GPT which includes data from billions of individual nodes finally allows developers like Azumo to build the type of bots we always envisioned. And once more recreate the voice of the company or publisher that our customers seek.
From customer care, to community management, remote medical assistance, legal advice, virtual assistance, tutoring, internet and enterprise search and more, will all be impacted.’
Look again at that middle paragraph. “And once more recreate the voice of the company or publisher that our customers seek.”
Re-create the voice they seek? Why not just say it? I’m serious. What are they talking about??
I’d like to ask our more objective GlobalData Thematic Analyst Daniel Clark (see above), “Daniel, if they’re putting human words in, and getting human-like words out, for humans to read, what’s the point? …
“No, wait, Daniel. On second thought, I’m a little afraid of your answer.”
(Dear Reader, I wouldn’t have believed him anyway.)
If you watch the CNN segment today it got 50 to 60% accuracy on medical license
Researchers have also tested ChatGPT on the United States Medical Licensing Exam (USMLE), which is a three-part exam that is required for medical licensure in the U.S. by all medical school graduates. They found that ChatGPT performed near or at the passing threshold for the three exams and demonstrated high levels of consistency and insight in its explanations.
The researchers concluded that large language models like ChatGPT have the potential to assist with medical education and decision-making. The paper states that clinicians within a virtual clinic called Ansible Health have already begun to experiment with ChatGPT in assisting with writing tasks, such as composing appeal letters and simplifying complicated medical reports to help patients better understand their conditions.
When I read about stuff like this, I'm glad I'm old. 😁