Nuclear Fusion - anticipating the aftershocks

The recording of last week's  webinar is now posted at

https://www.youtube.com/watch?v=kPJEEEQcCuc&t=3s

Another notable feature of ChatGPT 3.5 is its ability to handle longer conversations. In previous versions of GPT, the model was limited in its ability to sustain a conversation for an extended period of time. However, ChatGPT 3.5 has been designed to be more adept at maintaining a coherent and engaging conversation over an extended period. This makes it an ideal tool for chatbots and other applications that require sustained conversation with users.

One of the key advantages of ChatGPT 3.5 is its ability to adapt to the specific needs of different applications. The model is highly customizable, allowing developers to fine-tune it to fit the specific needs of their applications. This means that it can be used for a wide range of applications, from customer service chatbots to virtual assistants and beyond.

Overall, the introduction of ChatGPT 3.5 is an exciting development for the field of artificial intelligence. Its advanced conversation capabilities and ability to adapt to the specific needs of different applications make it a powerful tool that has the potential to revolutionize the way we interact with technology.

Here’s where it gets interesting folks. Everything you’ve read above this line was generated by ChatGPT in response to this request:

Can you write me a 300 word newsletter article on OpenAI and the introduction of ChatGPT 3.5

In less than five seconds, ChatGPT produced 304 words on OpenAI and the newest version of their natural language Chatbot. I could have continued to ‘converse’ with ChatGPT to make refinements to the newsletter article. For example, I could have asked ‘Can you expand on paragraph four and change the reading level to Grade 8’ or ‘make the newsletter article less formal and sassier’ and ChatGPT would have generated another option for me to consider.

ChatGPT 3.5 is a research release that was introduced to the world on November 29^th. It’s a large language model, natural language processing, conversational chatbot. ChatGPT 3.5 was trained on millions of human to human conversations and a corpus of information available on the internet up to 2021.

Adapted from the OpenAI website:

OpenAI employed Reinforcement Learning from Human Feedback (RLHF). OpenAI trained an initial model using supervised fine-tuning: human AI trainers provided conversations in which they played both sides - the user and an AI assistant. They gave the trainers access to model-written suggestions to help them compose their responses.

OpenAI created a reward model for reinforcement learning - they collected comparison data, which consisted of two or more model responses ranked by quality. They then took conversations that AI trainers had with the chatbot, randomly selected a model-written message, sampled several alternative completions, and had AI trainers rank them. Using these reward models, they fine-tuned the model using Proximal Policy Optimization and performed several iterations of this process.

The other key innovation underlying the Research Release of 3.5 is the provision of a user-friendly interface through which to ‘converse’ with the AI and the addition of the ability for everyday users to create an account to access ChatGPT 3.5.  It took Netflix 3.5 years to reach 1 Million users. It took Facebook 10 months. It took Spotify 5 months. Instagram took 2.5 months. As for ChatGPT? It took 5 days to reach 1 Million accounts.

The internet has been blowing up with stories about ChatGPT 3.5 and the astounding responses it generates. People are using the tool to write marketing plans, poems, legal documents, computer code, essays - almost anything that can be imagined. The results are polished, conversational, thorough, and important to note, sometimes incorrect. Critically, the model doesn’t ‘know’ what it’s talking about. It’s a statistical, probabilistic model that figures out (for a given prompt and based on its data set) what are the words most likely to satisfy the prompt and in what order they should flow.

But the fact that the model responds so quickly, fluidly, thoroughly, interactively, and conversationally appears almost magical and it’s raising important questions about what ChatGPT 3.5 can and cannot do. A brief experience with this technology has many wondering - what professional or personal realm will be safe from disruption?

Are you a writer? A marketer? A lawyer or paralegal? A student or a professor? An engineer or a software developer? As a consumer of information, are you prepared to be inundated by AI generated content?

ChatGPT 3.5, other platforms licensed by OpenAI and similar technologies like Sparrow from Google’s DeepMind will affect every walk of life. We are witnessing the dawn of a new era and because these foundational learning models are doubling in capability much, much more rapidly than Moore’s Law, the pace of change could be far faster than any of us have previously experienced. ChatGPT 4.0 is right around the corner and it has been trained on 500 times as many relationships. And the foundational technology of ChatGPT is already being built upon by 100’s of startups offering similar functionality, vertical market fine tuning, and new capabilities.

Have you heard of ChatGPT? Have you experienced it? What do you think? What opportunities do you see? What threats does this technology present? Are you prepared for your work to be disrupted and how can you get ahead of the wave?

The genie is out of the bottle. What do we wish for? What will we get? Let’s chat about the future of Artificial Intelligence and large language models!

Editor's note.  We are organizing a future webinar dedicated to exploring AI text processing technology.

AN UNSOLICITED RANT.  Customers just want a little respect. - received from Esir Pretne

I was raised to respect the customer as without one there was no reason for being in business.  The service ethic taught to me was that "the customer was always right!" Translated, that meant respect the customer even though they may be wrong. During my varied career in academia, government, industry, and even in the non-profit sector, the ethic served me well. But now that I am retired and have time to reflect. I am astounded with the lack of a service ethic today. Suppliers virtually everywhere act as if they have a monopoly, that the customer is a given, even going so far as to relegate service to no one - self service, or an AI device.  Oh, sure in marketing, the advertizing says otherwise; but in my experience service is a thing of the past.  Let me share a few personal examples and ask you if my experience is just due to my age and increasing crankiness.

First, let's look at the public service.  We here in Canada are blessed with public services from a sector that is horning in on seemingly everything: healthcare, education, transportation, to name a few. The socialization ideology is great - on the surface, but fraught with service issues that are always to be resolved by just a little more money from the taxpayer - the customer. More personnel. More jobs. You'd think that all government existed for is to create jobs - regardless of need and never never to wrap up program. But the waiting lists; the sustained pain continue think hip replacements where the waiting list is years, then there is passports and systems of gun control, payroll. The list goes on. As for public education, we are threatened with strikes - ignoring the impact on the customer, the parent, the child. As for the private sector we have the professions paid by the hour if not the minute.  There the "client" services reflects what was the origin of the word "client" in Roman times - one grade above a slave.  Of course inflation and supply shortages have only served to worsen service levels.

Now let me get down to some personal and practical examples.  Healthcare.  Three times in the last year my spouse has had to go to emergency with serious heart issues.  The wait times have been 14, 12, and 8 hours in emergency with dozens of others while the ambulance service's staff were waiting around until their patients got service.  Surely a waste of a precious source of supply. Then once admitted to emergency quarters, there was an additional wait as staff attended to the higher priority of entering data into systems that don't communicate.  We had to provide personal information three times in succession during registration because people a few feet from one another couldn't get what we had previously been entered into the million dollar systems.

Then there are the research universities with their high paid tenured professionals. From time to time I have asked for their time so we could help off-campus, community-minded customers learn what the research university knew or was studying.  The almost universal response has been "Too busy". I interpret that to mean, "I have higher priorities - than you, the community".  I was once asked by a research university what criteria might be introduced to help bridge the academic/community gap. None of those I provided were viewed as possible where research seemed to be the only priority.  Even students complained that research staff were unavailable for tutouring or reluctant teachers. The priorities of a monopoly are supply management.  There is no threat that the customers might go elsewhere or complain to the funder - often the government or that complaining would make any difference anyway.  What to do?

Market regulators are notoriously overmatched. Once a supplier gets and enjoys the privilege of a monopoly - think healthcare and education, those in power - unions including the professions. Amazon and Microsoft, telecommunications and airlines, go to extraordinary lengths to protect the status quo and resist pressures to transition. It's not so much that power corrupts, but voluntary pivoting is rare - it is painful.

When the supplier gains overwhelming power over the consumer as in a public service monopoly, when market forces no longer work to govern standards, truth, and costs; it is time for the market regulator to intervene. Yes, even in a free market, a regulator is warranted for mitigating supply/demand, supplier/consumer excesses. But in public services the operator is the regulator, there is no source of influential consumer appeal...just promises of reform and appeals to adhere to the intractable ideology.

As I recall from my consulting days, public service monopolies including not-for-profit lobbyists, spend time, money, and sometimes dedicated legal resources to protect their mutual interest. Of course corporations also have their government influential peddlers - often public sector retirees. Where in public service negotiations, budget development, and service standards assessment is the consumer represented? Where there is no consumer rep, supplier interests prevail. When is enough enough?

The public health system is notorious in resisting innovation and competition is nil. It could take a lesson from small hospitality services where customer respect is still of benefit to the server in the form of tips. (God I hate the new payment devices that impose tip options.) Comparing Canada's public health with private service delivery triggers more ideology and conjures up horrific stories about the US before a constructive dialogue can even begin and/or discuss meaningful comparisons. Reform seems hopeless. Just more money! Could personalized technology dedicated to knowing and serving consumer interests help? Think innovation such as precision medicine through genetic engineering and cellular therapies in healthcare. Think Uber and Airbnb as the private sector capitalizes on surplus supplier assets. Most technology development is supplier driven by an interest in consumer service  whether needed, wanted, desired or not. Open source innovations promise more customer respect as an exception. Gotta love "open source".  

Service reform would be a huge reset starting with restoration of respect for the consumer, the patient, the client. More smiles. Less waiting. And a lot more competition.

There - I've got it off my chest.  I suspect I am not alone.

Signup for Troy Media's daily editorial and news updates HERE

Solving labour shortages through innovative inclusion. Contact us HERE