Human-in-the-Loop Artificial Intelligence

Teams of humans and AI agents will be the model for the organization of the future.

There are many tasks which Artificial Intelligence (AI) can’t perform well without help and supervision; it has an unfortunate tendency to hallucinate (make up an answer). For that matter, there are many tasks humans can’t perform well without help or supervision and humans are expensive and in short supply, at least in the work force. Two techniques have been developed in the last year of experimentation with AI and Large Language Models (LLMs) like ChatGPT: multi-agent collaboration and human-in-the-loop processes. Both help mitigate hallucination and produce content faster and more cheaply than humans can alone.

Multi-Agent Collaboration

An agent is nothing more than a packaged use of an LLM. The simplest agents accept a human prompt, ask AI to respond, and return the result. More complex agents retrieve information the LLM has not been trained on from a database or from a web search and send that information along with the request to the LLM. Agents can ask the LLM to write code in order to generate a response to the human prompt. For example, if the human wants a graph of the Dow Jones Average (DJA) for the next month compared to the S&P, several agents might be involved. The LLM will not have been trained on data as recent as last month, and so one agent retrieves the daily closing prices of both the DJA and the S&P from the web using the general knowledge of the LLM to formulate the web requests. A second agent writes computer code necessary to produce a graph of the two indices on the same set of axes. A third agent executes the code and the resulting graph is displayed to the user.

Using multiple agents is also a good way to reduce hallucination. Agent One (the writer) generates a draft answer to a question; Agent Two (the fact-checker) does web searches to validate assertions in the draft; Agent Three is a reflection agent which critiques the draft both using the fact-checking results and looking for bad style or bad usage. The answer then goes back to Agent One for revision and back though the gauntlet of agents Two and Three until all are satisfied. The final answer is sent to the human who asked the question. Interestingly (for technical reasons I won’t explain here) even if all agents use the same LLM, they can disagree constructively. However, using different LLMs for the different agents makes the whole process even more robust and reliable.

Human-in-the-Loop

Suppose the writer, fact-checker, and reflection agents above never reach agreement. Drafts could go round and round forever chewing up expensive LLM cycles. One solution is an arbitrator agent. Another solution is to add a human into the loop. For example, after the first draft has been written, fact-checked, and critiqued, a human is given an opportunity to redraft herself, to modify or accept the critique, and to arbitrate the fact-checking. Depending on what the human decides, the article could go another round and then come back to the human again. The process ends when the human accepts a draft, perhaps after making some more modifications. In more complex processes, humans can be involved in any step the application designer feels is helpful.

A Demonstration

I programmed a demonstration application to write a news story using either the minutes or a transcript of a meeting as a source. With the shrinkage of local news outlets, there are never enough reporters to attend all the public meetings which affect our lives and should influence our votes. AI and LLMs alone can’t create a well-written news story accurately enough. But human editors working with AI can turn out stories which are well-written and accurate… and cheap enough for a local newsroom to afford.

First the human editor specifies how many words the article should be and whether the source is on the internet or the editor’s computer using the screen below.

In the next screen (not shown) the editor gives a URL or selects a local file as the source. An input agent obtains the source document and extracts text from it. The writer agent uses the transcript or minutes to draft an article. The reflection agent does double duty: fact-checking against the source document and checking style. The human editor can then edit either the article or the critique or accept the article as written using the dialog pictured below. If there is a human or agent critique, the cycle repeats with a rewrite.

The source document I used for this article was the draft minutes of a Stowe Select Board meeting. The reflection agent criticized the first draft for not starting out with the most important facts (“burying the lede” in news jargon), for not clearly explaining the disagreement among the Board Members, and from being sloppy with some facts. The second draft was better in these respects and was accepted by the reflection agent. However, I (playing editor) felt it was still not clear in one respect and wrote my own critique. The third draft passed muster. The whole dialog is down below but it’s long.

You can play with this demo app at https://meeting-reporter.streamlit.app/. I don’t charge anything for using it, but you do need to provide your own OpenAI API key to get past the first screen because the LLM being used is ChatGPT4. OpenAI will only charge you a few cents per use, but you do need an account and a paid API Key. If you don’t know what an API key is, you are like most of the world and you don’t want or need one, at least not yet.

Update: ChatGPT-4o is now free to everyone and include access to custom GPTs, a kind of AI app. I've created a version of human-in-the-loop meeting reporter as a custom GPT which you free without either an API key or a paid subscription to anything https://chatgpt.com/g/g-roNR24Ty6-collaborative-meeting-reporter.

Note to nerds: if you want to see or play with the source code, it is open source and available at https://github.com/tevslin/meeting-reporter. It uses Langgraph and Streamlit libraries with a tkinter version available and is hosted in the Streamlit cloud. The LLM is ChatGPT4-Turbo.

See also:

An AI Debate for another example of multi-agent collaboration

https://blog.tomevslin.com/ai/ for more on AI in general

The Transcript of the Session (tl;dr)

The original draft

 In a special meeting held on Wednesday, March 29, 2024, the Stowe Selectboard, led by Chair Billy Adams, convened to address the urgent matter of the Stowe Short-Term Rental Registry Ordinance. The meeting saw attendance from both Selectboard members and a significant number of the public, with 26 individuals participating via Zoom. The session was marked by discussions on the appropriateness of the meeting's timing and the methods for deciding on the rescindment of the recently adopted ordinance.

The Selectboard approved the agenda unanimously after a brief discussion on whether the meeting was necessary ahead of the regular April 10 meeting. The urgency was attributed to a petition received on March 28, necessitating a special meeting to adhere to statutory requirements for public notice. Despite some members expressing concerns over the rushed nature of the decision-making process, the board agreed to proceed with planning for a special meeting of the voters.

Set for Wednesday, May 1, 2024, at the Stowe High School Auditorium, the special meeting aims to vote on whether to rescind the town’s Short-Term Rental Registry Ordinance, adopted on February 29, 2024. The decision came after confirming the venue’s availability and the necessity of holding the meeting within a 60-day timeframe as required by law. Concerns were raised about ensuring adequate participation, with suggestions for a later start time to accommodate working residents and considerations for childcare to aid attending parents.

Amidst the technical discussions, a deeper undercurrent of division within the Selectboard was evident, with members voicing the need for a more collaborative approach moving forward. The meeting concluded with a call for unity and a more deliberate approach to decision-making, highlighting the tensions surrounding the short-term rental ordinance debate.

The Selectboard ultimately decided against holding an informational meeting but agreed on the creation of a fact sheet to be made available online and in the newspaper, aimed at clarifying the ordinance details and the upcoming special meeting process.

Critique from the reflection agent

The article is well-written and adheres to the facts of the source material. However, there are a few areas where clarification would improve comprehension. Firstly, the ordinance the Selectboard is considering rescinding is introduced too late in the article. Mention it earlier to provide necessary context. Secondly, the "deeper undercurrent of division" within the Selectboard needs more explanation. What are the differing views? Lastly, the final paragraph states the Selectboard decided against an informational meeting. This seems contradictory to the fact sheet creation. Clarify this decision-making process.

Second draft

In a special meeting convened on Wednesday, March 29, 2024, the Stowe Selectboard, under the leadership of Chair Billy Adams, gathered to deliberate on the pressing issue of possibly rescinding the Stowe Short-Term Rental Registry Ordinance, adopted on February 29, 2024. The meeting, attended by Selectboard members and the public, with 26 individuals joining via Zoom, was characterized by significant discussions focusing on the timing of the meeting and the procedural aspects of revoking the recently instituted ordinance.

From the outset, the urgency of the meeting was made evident, with the Selectboard unanimously approving the agenda. This urgency was driven by a petition received on March 28, compelling the need for a special meeting to comply with the legal requirements for public notification. Despite reservations voiced by some members regarding the expedited nature of this process, there was consensus on moving forward with the planning of a special voter meeting.

Scheduled for Wednesday, May 1, 2024, at the Stowe High School Auditorium, this special meeting is aimed at deciding whether to overturn the town's Short-Term Rental Registry Ordinance. This decision was reached after affirming the venue's availability and recognizing the legal obligation to conduct the meeting within a specific 60-day window. The discussion also touched upon ensuring broad participation, with proposals for a later starting time to accommodate residents working out of town and the provision of childcare services to support attending parents.

Amid these logistical and procedural deliberations, a palpable division within the Selectboard came to the fore, underscoring differing perspectives on the approach to the short-term rental ordinance debate. This internal discord prompted calls from members for a more unified and thoughtful decision-making process going forward.

Concluding the meeting, the Selectboard opted against hosting an informational meeting, instead agreeing on disseminating a fact sheet through online platforms and local newspapers. This approach aims to demystify the specifics of the ordinance and the procedural details of the forthcoming special meeting, addressing the public's need for clear and accessible information.

My Critique

The minutes do not say that they "agreed on disseminating a fact sheet through online platforms and local newspapers. They simply say that Mr. Adams asked Mr. Safford to have the staff prepare a fact sheet. Please be precise.

Final Article

Stowe Selectboard Schedules Special Meeting Amidst Division

11/04/2024

In a special meeting convened on Wednesday, March 29, 2024, the Stowe Selectboard, under the leadership of Chair Billy Adams, gathered to deliberate on the pressing issue of possibly rescinding the Stowe Short-Term Rental Registry Ordinance, adopted on February 29, 2024. The meeting, attended by Selectboard members and the public, with 26 individuals joining via Zoom, was characterized by significant discussions focusing on the timing of the meeting and the procedural aspects of revoking the recently instituted ordinance.

From the outset, the urgency of the meeting was made evident, with the Selectboard unanimously approving the agenda. This urgency was driven by a petition received on March 28, compelling the need for a special meeting to comply with the legal requirements for public notification. Despite reservations voiced by some members regarding the expedited nature of this process, there was consensus on moving forward with the planning of a special voter meeting.

Scheduled for Wednesday, May 1, 2024, at the Stowe High School Auditorium, this special meeting is aimed at deciding whether to overturn the town's Short-Term Rental Registry Ordinance. This decision was reached after affirming the venue's availability and recognizing the legal obligation to conduct the meeting within a specific 60-day window. The discussion also touched upon ensuring broad participation, with proposals for a later starting time to accommodate residents working out of town and the provision of childcare services to support attending parents.

Amid these logistical and procedural deliberations, a palpable division within the Selectboard came to the fore, underscoring differing perspectives on the approach to the short-term rental ordinance debate. This internal discord prompted calls from members for a more unified and thoughtful decision-making process going forward.

Concluding the meeting, Chair Adams specifically requested Mr. Safford to have the staff prepare a fact sheet aimed at clarifying the specifics of the ordinance and the procedural details of the forthcoming special meeting. This focused effort reflects the Selectboard's intention to address the public's need for clear and accessible information without definitively concluding on the method of dissemination.

Live on WDEV - Remembering Tim Hayward

Today’s Common Sense call-in radio show with Bill Sayre on WDEV will be “Remembering Tim Hayward—his life and his work, his immense contribution (though out of the lime light)  to making Vermont a better, more affordable, place to live, work and raise a family.” The show is at 11AM on FM 96.1, 96.5, 101.9, and AM 550 and Live Streaming from https://www.radio.net/s/wdevfm.  I’ll be Bill’s guest but all are more than welcome to call in with memories of Tim: 802 244 1777.

See also:

In Memoriam: Timothy Y. Hayward

In Memoriam: Timothy Y. Hayward

My best friend Tim Hayward died early Sunday morning. Over the last 50 years, he has been the greatest force for good and effective government in Vermont that you never heard of. He has done more for Vermont and all of us here than most people you hear about daily. Governor Phil Scott’s statement this morning gives the highlights:

“I join so many Vermonters in mourning the passing of our friend Tim Hayward. Tim was a true giant of quiet, selfless service, and he worked every single day to do his part and then some. 

“He was a proud marine, legislator, and senior advisor to many – including Governor Snelling, Congressman Smith, Congressman Mallory, Senator Jeffords – and for 8 years he served as Chief of Staff for Governor Douglas.

“Tim’s most important acts of service, however, were not the many senior positions he held, or officials he advised. His very best work – work he clearly loved – was his mentorship to so many other public servants, who were taught by Tim and who benefited from his example. His impact on our state does not end with him, but will live on for generations to come.

“I was so fortunate that, after being elected Governor myself, Tim agreed to lead the transition team. He helped lay a strong foundation in this Administration.  And he prepared my team for the road ahead.  Always with the rigor and discipline of an old marine, the wisdom of a great coach and the principled heart of the servant citizen.

“I offer my sincerest condolences to his wife Susan, his three children Nathanial, Heidi and Zak, and all of his beloved family and friends.

“In honor of his legacy and service, I have ordered the Vermont State Flag to fly at half-staff on the day of his services.”

Tim was the behind-the-scenes detail person in both the Snelling and Douglas administrations. He was the master of constructive principled compromise. He was the frank and honest voice that political leaders usually don’t hear.

Most people in politics gravitate to wherever a camera is pointing; Tim assiduously hid from attention. It was never about him.

Both Snelling and Douglas had more national influence than you would think a Vermont governor would have, largely through their leadership positions in the National and Republican Governors Associations. Their influence, which was both good for Vermont and good for the country, owed much to the respect that Tim earned from the staff of these organizations and the staff of other governors (and Canadian premiers).

I had the good fortune to work with Tim during parts of both the Snelling and the Douglas Administration. But it wasn’t just luck that got me there. Tim decided that there were jobs that I would be helpful at, convinced the governors to offer them to me, and told me I had no choice but to take a furlough from the private sector. Once I was in the positions, he did his best to help me do what needed doing.

We hiked the Worcester and Manfield ranges with three generations of dogs. I miss Tim Hayward terribly; so will Vermont.

Tim would want me to thank the staff at Central Vermont Medical Center in Berlin, Vt who did a wonderful and caring job through the ordeal of the marine’s last battle for both him and his family. Thank you.

How To Get Useful Answers with ChatGPT Plus

Chatbots powered by AI and large language models (LLMs) don’t always answer questions accurately. Neither do humans for that matter. We can’t blindly trust what ChatGPT tells us nor what we read on Facebook or X. Not everything we learned in school is correct. Your neighbor may be smart but she’s not always right. Even mommies make mistakes. Reporters have both biases and blind spots. So how do we get “the truth” or at least useful answers?

ChatGPT almost never gives links to support its answers because LLMs like ChatGPT don’t “know” how they know things. We don’t remember how we know most of what we know. Do you remember when you learned each word in your vocabulary? When and where did you learn that the sun rises in the East? LLMs are trained by reading lots of stuff, some contradictory, some nonsense, much biased. There is inevitably some bias in the selection of training material just as there was bias in the education you got. They often get “finishing school” training from humans who impart their own biases directly (witness Google’s recent embarrassment at what was supposed to be politically correct image generation which ended up deciding that US history is so unracist that the “founding fathers” were actually mostly Indians and women and that Nazi soldiers and Vikings were mainly non-Caucasian). If the answer to a question is important to you, you want to know the source of the information from which the answer was generated.

However, because ChatGPT Plus, which costs $20/month, has access to Bing browsing (which the free version of ChatGPT does not Update: The free version of GPT-4o now has browser access as well), it can find links to support its answers if you tell it to. Sometimes, if properly prompted, it’ll even find links which cast doubt on its assertion or reveal uncertainty. A simple way to get confidence in what ChatGPT tells you is to ask for supporting links. For example, I asked:

“Is volcanic activity a significant contributor to climate change?”

The first answer I got (don’t feel you have to read it all) was:

Lots of assertions but no backup for them. They could even be hallucinations. Moreover, the last paragraph is politically correct (and perhaps factually correct) and almost certainly a result of the finishing school training ChatGPT had.

So I said:

“Please provide links relevant to the assertions in the last answer and explain how these links support, qualify, or contradict the assertions. for each link. Show the title of the page linked to.”

And got back an answer with references and links (the blue quotes in brackets. They are not live in the picture below but you can click here to open a browser window on my shared chat and the links will be live):

The references appear to be credible and, on following the links, I see that they do support what ChatGPT says they support. However, the evidence appears one-sided and I already suspect ChatGPT of being partial to accenting the contribution of anthropogenic climate change. To assure I’m seeing beyond this possible bias, I asked:

“What references are there including links which would appear to disagree with these conclusions?”

Interesting response (click here for live links):

Now we have pretty good information on vulcanism and climate and also know some of the questions which remain to be definitively answered.

This isn’t only a way to check what an LLM tells you. The first answer could have been something someone posted on Facebook, an essay somewhere, something someone told you in a bar, or a news article. You can send the text or a link or upload a document to ChatGPT Plus (Update: You can now do the with free version of GPT-4o as well) and ask it to generate references supporting and questioning assertions. Used in this way LLMs become a tool for detecting misinformation no matter what the source and getting reasonably reliable answers rather than yet one more way to generate half-truths and propaganda.

Practice in checking references is a crucial part of teaching critical thinking. Critical thinking and validating information are now much more valuable skills than a bunch of memorized facts. This use of LLMs must be taught to anyone old enough to go online. Within a year, use of LLMs and browsing will be indistinguishable because AI is being built into search at warp speed and LLM’s with search are better sources by far than classic Googling (open question on whether Google itself survives this transition). Every school and every organization must teach lessons in LLM use like the one in this post now. Avoiding LLMs because they can be a source of misinformation rather than embracing their proper use is educational and managerial malpractice.

See also:

AI Can Help Deal with Misinformation (for a GPT designed to help with fact checking)

Why You Want to Use Free ChatGPT-4o Instead of Search

More posts on AI

Three Toxic Phrases

“You’re not needed.”

Terrible thing to tell a person. We’re descended from tribal ancestors as far back as when we were swinging from trees. Anyone in a tribe who “isn’t needed” is well on the way to being out of the tribe. A tribal animal without a tribe usually ends up dead. Not being needed makes us anxious, to say the least.

But “you’re not needed” is the message we send to people when we tell them that we’ll feed them and clothe them and house them whether or not they “choose” to work.  The message isn’t only morally and ethically wrong, it’s also incorrect. We do need the contribution of everyone who can work even with automation. We have an aging population (including me) who will need more and more care and be able to contribute less and less. Our infrastructure is decrepit and needs to be revamped to deal with the effects of climate change. We’re short of housing. Population has peaked in almost all developed countries and is on the verge of a worldwide decline. We need all hands on deck. There’s no good reason to signal to anyone that they’re not needed.

One of the pathologies that afflict people who have been told they are “unneeded” is drug abuse. The difficulty or recovering from addiction has been compounded by another terrible message:

“You’re not responsible for your addiction.”

Sounds like a nice thing to say but it isn’t. It says you are helpless. It says you have no agency. It says you are dependent on others.

We know from the long experience of Alcoholics Anonymous that only people who take responsibility for their addiction can be cured. Saying “I’m Tom and I’m an alcoholic” has been a first and necessary step to a cure for many Toms. According to the signs sponsored on the windows of Burlington Airport by a proliferation of drug-treatment nonprofits, there isn’t even such a thing as drug abuse; people have “substance use disorder”. Apparently their treatment should start by saying “I’m Tom and it’s not my fault. I can’t do anything about my addiction. I need a safe place to take my drugs.” It is incredibly hard to break an addiction; we make it harder by telling people that they are not in control of their own lives.

“You’re not from our tribe.”

Because tribalism is in our DNA, it’s easily resurrected by demagogues. The often-successful melting pot which is the USA is being stirred into a toxic caldron by reawakening old tribalisms and inventing new ones.  The endless conflicts in the Middle East and the ignored but just as deadly wars in Africa come from tribalism (sometimes reinforced by religion). We’re not Americans first and then possibly an ethnic heritage (or two or three); we’re ethnic first and not so sure we want to be Americans. We’re not Americans who try to figure which people we want to govern us; we’re Progressives, Democrats, Republicans, MAGAs trying to assure that our minority rules. We’re not people; we’re cis-gendered men, women, L,G,B,T,Q, and/or+. Our civil rights are not the ones the creator endowed all people with, they come from our tribes and are at the expense of other tribes. This doesn’t work.

Charmed Phrases

“We need your work.” (we’ll help you find some)

“Only you can cure your addiction.” (we’ll help you if you want to be cured).

“E Pluribus Unum.”

An AI Debate

Resolved: Vermont should devote available limited resources only to mitigating the effects of climate change rather than attempting to prevent it.

The debate far below is between AI “agents” This post is about a positive use of AI and suggestions for further use. The debate is just an example. But the proposition debated here interests me as well. I posted some on the substance here and will post more in the future.

An AI agent (sometimes called an assistant) is an interface to a large language model (like ChatGPT) bundled with instructions giving it a special role and optionally tools and data which help it perform that role. Agents collaborate with each other and optionally with humans to perform a task AI (often in the form of a chat manager agent) guides the collaboration and decides when the task has been accomplished. These agents are all getting their intelligence from ChatGPT. Using a tool called AutoBuilder from Microsoft, I instructed ChatGPT to create a team of agents including a moderator and a judge capable of debating any proposition given to them. It took a couple of hours for me to get it working right since I had to learn to use AutoBuilder, which is not quite ready for use by non-nerdy people.

The debate illustrates how quickly and easily tools can be built with AI. The debate format is a useful way to quickly get both sides of an argument as well as some of the more obvious rebuttals. When I do post more on the substance of whether Vermont should devote its resources to preventing climate change rather than to mitigation, I’ll use the debate as a checklist to see if I’ve missed arguments or obvious flaws in arguments. It would be more helpful if I had the agents do research and present links but that’s an enhancement for another day.

This tool could serve as the basis for debate training or debate practice if humans rather than agents played the role of one or more of the debaters.

Having AI argue with itself is a way to counter the tendency of large language models like ChatGPT from making up plausible but hallucinatory answers to questions. Bot checking bot would be even more powerful if different LLMs were behind the various agents, which could be done relatively easily. This quality control on LLM output the same way that inspectors inspect (or should inspect) factory output. The Fact or Fiction GPT I posted about here is an example of a bot being used as quality control on human writing.

Collaborative teams of agents and humans are part of the immediate future of how many people will work. The debate is just an example. There could be, for example, a product team with marker-researcher, designer, programmer (agents can write code!), tester, and social media agents. Humans can play some of these roles and have input into all of them if the collaboration is so designed.

If you’d like to have the toy I built debate a subject of your choice, you can so here. It’s dead simple to use and requires no installation; but the catch is you have be enough into AI to have a paid API key from OpenAI. If you don’t know what that is, you don’t want to get one just to challenge the debate team.

If you are a nerd and want to build your won debate team, enhance the one I built, or just see how the AutoBuilder tool works, it’s all open source at https://github.com/tevslin/debate_team.

The generated debate is below.

*********************************************************************************************************************************************************************************************************

affirmative_constructive_speaker (to chat_manager):

As a constructive affirmative debater, I will present arguments in favor of the proposition that Vermont should devote its available limited resources to mitigating the effects of climate change rather than attempting to prevent it.

**1. Proportionality of Impact:**

Vermont, as a small state, contributes a relatively minuscule amount to global greenhouse gas emissions. Even if Vermont were to invest heavily in prevention, the global impact would be negligible. Therefore, it is more pragmatic for Vermont to focus on adaptation strategies that protect its citizens and infrastructure from the inevitable effects of climate change, such as extreme weather events, rather than attempting to significantly alter global emission trajectories.

**2. Immediate Benefits:**

Mitigation efforts provide immediate and tangible benefits to the local population. By investing in infrastructure that can withstand extreme weather, improving emergency response systems, and enhancing the resilience of agricultural practices, Vermont can ensure that its residents are better protected against the current and increasing impacts of climate change.

**3. Economic Prudence:**

The economic argument for focusing on mitigation is strong. The costs associated with climate-related disasters can be astronomical, and by investing in mitigation now, Vermont can avoid much larger expenses in the future. This is not only fiscally responsible but also ensures that the state's limited resources are used in a way that provides the most benefit to its residents.

**4. Social Equity:**

Mitigation efforts can be designed to prioritize the most vulnerable communities in Vermont, which are often the hardest hit by climate change. By focusing on adaptation, the state can address social inequities and ensure that all Vermonters have the resources and support they need to cope with the effects of climate change.

**5. Encouraging Innovation:**

Investing in mitigation can spur innovation in areas such as renewable energy, sustainable agriculture, and resilient construction practices. This not only helps Vermont deal with the effects of climate change but also positions the state as a leader in developing technologies and strategies that can be exported and implemented elsewhere.

**6. Global Responsibility:**

While prevention is a global responsibility, mitigation is a local necessity. Vermont can act as a model for other states and regions by demonstrating how to effectively adapt to climate change, thereby fulfilling its global responsibility through leadership and example.

In conclusion, given Vermont's limited resources and the broader context of global emissions, it is both strategic and necessary for the state to prioritize mitigation of the effects of climate change. This approach ensures the well-being of its residents, the sustainability of its economy, and the resilience of its communities.

--------------------------------------------------------------------------------

negative_construction_speaker (to chat_manager):

As the negative constructive speaker, I will present arguments opposing the proposition that Vermont should devote its available limited resources only to mitigating the effects of climate change rather than attempting to prevent it.

**1. Ethical Responsibility:**

Vermont, like all states, has an ethical responsibility to contribute to the prevention of climate change, not merely to adapt to its consequences. By focusing solely on mitigation, Vermont would be neglecting its duty to future generations who will face the compounded effects of today's inaction. Prevention is a shared global burden, and every entity, regardless of size, must partake in the collective effort to reduce emissions.

**2. Long-Term Strategic Investment:**

Investing in prevention is a long-term strategic investment. While mitigation addresses the symptoms of climate change, prevention tackles the root cause. By investing in renewable energy, energy efficiency, and other preventative measures, Vermont can help slow the progression of climate change, ultimately reducing the severity and cost of future mitigation efforts.

**3. Economic Opportunities:**

Prevention efforts can also drive economic growth and job creation. Developing a green economy through investments in renewable energy and energy efficiency technologies can create new industries and employment opportunities. This can diversify Vermont's economy and make it more resilient to the economic disruptions that climate change may bring.

**4. Leveraging Influence:**

Vermont's commitment to prevention can have an outsized influence beyond its borders. By setting ambitious targets for emission reductions and implementing progressive environmental policies, Vermont can inspire other states and countries to follow suit. Leadership in prevention can enhance Vermont's reputation and influence in national and international discussions on climate policy.

**5. Risk Management:**

Relying solely on mitigation is a risky strategy that assumes the state can adequately adapt to all future climate-related challenges. However, the pace and scale of climate change may outstrip Vermont's ability to respond, leading to irreversible damage. Prevention reduces this risk by aiming to limit the severity of climate change itself.

**6. Moral Leadership:**

Vermont has a history of environmental leadership and moral authority on progressive issues. By investing in prevention, Vermont can continue to lead by example, demonstrating that even small states can make meaningful contributions to a global challenge. This moral leadership can galvanize action and support for climate initiatives within and beyond its borders.

In conclusion, while mitigation is undoubtedly important, it is not sufficient on its own. Vermont should balance its efforts between mitigation and prevention to address both the immediate and long-term challenges of climate change. This balanced approach is not only more ethical and strategic but also offers economic benefits and aligns with Vermont's legacy of environmental stewardship and moral leadership.

--------------------------------------------------------------------------------

affirmative_rebuttal_speaker (to chat_manager):

As the affirmative rebuttal speaker, I will now address the points made by the negative team and reinforce the proposition that Vermont should focus its limited resources on mitigating the effects of climate change rather than attempting to prevent it.

**1. Ethical Responsibility:**

The negative team argues that Vermont has an ethical responsibility to contribute to the prevention of climate change. While this is a noble sentiment, it overlooks the practical limitations of Vermont's influence on global emissions. Ethical responsibility must be balanced with pragmatic action. Vermont's ethical duty extends to protecting its own citizens from the immediate threats posed by climate change, which is best achieved through targeted mitigation efforts.

**2. Long-Term Strategic Investment:**

The negative team suggests that prevention is a long-term strategic investment. However, the urgency of climate change requires immediate action. Mitigation is not merely addressing symptoms; it is a proactive approach to safeguarding Vermont's communities right now. Furthermore, the long-term benefits of prevention are uncertain, given the state's small contribution to global emissions, whereas the benefits of mitigation are concrete and immediate.

**3. Economic Opportunities:**

While the negative team points to economic opportunities in prevention, it is important to note that mitigation efforts also drive economic growth. For example, investments in resilient infrastructure can create jobs in construction, engineering, and urban planning. Moreover, mitigation can stimulate innovation in sectors directly relevant to Vermont's needs, such as sustainable forestry and agriculture.

**4. Leveraging Influence:**

The argument that Vermont's commitment to prevention can influence others is speculative. Vermont's true influence lies in its ability to showcase successful adaptation strategies. By becoming a living laboratory for effective mitigation, Vermont can provide a replicable model for others to follow, which is a more tangible form of influence than setting ambitious but potentially unattainable emission reduction targets.

**5. Risk Management:**

The negative team's claim that relying solely on mitigation is risky fails to recognize that prevention efforts are equally, if not more, uncertain. The pace of global emissions reduction is slow and unpredictable, whereas the impacts of climate change are immediate and escalating. Vermont must manage the risks it faces today, which are best mitigated through local action.

**6. Moral Leadership:**

Finally, the negative team's appeal to Vermont's moral leadership overlooks the fact that leadership can take many forms. Vermont can lead morally and effectively by demonstrating how a state can adapt to and thrive in the face of climate change. This is a powerful form of leadership that resonates with the immediate needs and concerns of people around the world.

In summary, while prevention is an important aspect of the global response to climate change, Vermont's limited resources must be allocated to where they will have the most significant impact. Mitigation offers immediate protection, economic benefits, and a practical form of leadership that aligns with Vermont's capacity and the urgent needs of its citizens.

--------------------------------------------------------------------------------

negative_rebuttal_speaker (to chat_manager):

As the negative rebuttal speaker, I will now address the points made by the affirmative team and reinforce the stance that Vermont should not solely focus on mitigating the effects of climate change but also invest in prevention.

**1. Proportionality of Impact:**

The affirmative team argues that Vermont's contribution to global emissions is minuscule and therefore should focus on mitigation. This argument fails to consider the cumulative effect of every small contribution. If every small state or entity adopts this mindset, the collective inaction would be significant. Vermont's efforts in prevention, though seemingly small, are part of a global tapestry of climate action necessary to effect change.

**2. Immediate Benefits:**

While mitigation provides immediate benefits, prevention has the potential to reduce the need for such measures in the long term. By investing in prevention, Vermont can help slow the rate of climate change, which may reduce the frequency and severity of extreme weather events, ultimately benefiting the local population in a more sustainable and enduring way.

**3. Economic Prudence:**

The affirmative team's economic argument for mitigation overlooks the long-term economic benefits of prevention. Investing in renewable energy and energy efficiency not only contributes to emission reductions but also can lead to energy independence, lower energy costs, and a more sustainable economy for Vermont in the long run.

**4. Social Equity:**

The affirmative team suggests that mitigation efforts can prioritize vulnerable communities. However, prevention can also be designed with social equity in mind. For example, energy efficiency programs can lower utility bills for low-income households, and renewable energy projects can provide jobs and investment in disadvantaged areas.

**5. Encouraging Innovation:**

While mitigation can spur innovation, so can prevention. Vermont's investment in prevention technologies can lead to advancements in renewable energy and energy efficiency, which are critical sectors for the future. These innovations can then be shared globally, contributing to the prevention of climate change on a larger scale.

**6. Global Responsibility:**

The affirmative team posits that mitigation is a local necessity while prevention is a global responsibility. However, this dichotomy is false. Global responsibility includes local action in prevention. Vermont can and should contribute to both global prevention efforts and local mitigation strategies, as they are not mutually exclusive.

In conclusion, the affirmative team's arguments for focusing solely on mitigation do not adequately address the importance and benefits of prevention. Vermont should strive for a balanced approach that includes both mitigation and prevention, ensuring the state fulfills its ethical obligations, maximizes long-term economic benefits, and contributes to global efforts to combat climate change.

--------------------------------------------------------------------------------

debate_judge (to chat_manager):

**Constructive Affirmative Speaker:**

Relevance: 5

Evidence: 4

Effectiveness: 4

Feedback: The speaker provided a well-structured argument with clear points supporting the proposition. The emphasis on the proportionality of impact and immediate benefits was compelling. However, the argument could be strengthened with more specific data on Vermont's emissions and examples of successful mitigation strategies already in place.

**Negative Constructive Speaker:**

Relevance: 5

Evidence: 4

Effectiveness: 4

Feedback: The speaker effectively countered the proposition with a strong ethical stance and a focus on long-term strategic investment. The economic opportunities and risk management points were particularly persuasive. To improve, the speaker could benefit from incorporating more Vermont-specific examples of how prevention efforts have led to economic growth.

**Affirmative Rebuttal Speaker:**

Relevance: 5

Evidence: 3

Effectiveness: 4

Feedback: The rebuttal was on point, directly addressing the negative team's arguments. The speaker effectively argued the urgency of action and the practicality of mitigation. However, the rebuttal would be more convincing with additional evidence supporting the immediate benefits of mitigation over the long-term uncertainty of prevention.

**Negative Rebuttal Speaker:**

Relevance: 5

Evidence: 4

Effectiveness: 4

Feedback: The speaker provided a strong rebuttal to the affirmative's points, emphasizing the importance of a cumulative approach to prevention. The arguments about the long-term economic benefits and social equity were well-articulated. To enhance the argument, the speaker could include more information on how other small states or entities have successfully influenced global climate action through prevention efforts

I’m Gonna Be MAGA-Canceled

Trump threatens Haley supporters.

According to The New York Times, Donald Trump posted that anyone who donates to Nikki Haley’s campaign “from this moment forth will be permanently barred from the MAGA camp.”

I contributed to her previously but apparently could have been forgiven. Just in case – and because I hope she stays in the race, I contributed again today. #hangintherehaley

Frankly I don’t like being threatened. Trump’s prediction in 2020 that there might be violence if he weren’t reelected (made before he actually wasn’t reelected) cemented my decision not to vote for him then.

We’ve had less than a week since the Republican primary became a two-person race. It’s way too soon to call it over. Just for a maybe, if Biden starts to rise in the polls because of a strengthening economy and/or more signs of senility from Trump, Republican voters in later primaries may start thinking long and hard about whom they are going to get for four more years if they nominate Trump.

BTW I’m not canceling anyone, even people with MAGA hats; they’re always welcome to change their minds and are entitled to their opinion and their vote even if they don’t.

See also:

The Trump-MAGA Purge (WSJ)

Cheney Says Haley Should Stay in G.O.P. Primary Through Super Tuesday (NYT)

English is Now the Most Powerful Programming Language

We better teach kids how to use it.

“What’s the most powerful programming language?” I asked grandson Jack as we zoomed together on our latest project.

“ChatGPT?” he guessed, perhaps giving some deference to my known obsession.

“Close,” I said. Then, parroting a recent internet meme, I told him the answer: “English because you use it to tell ChatGPT what program you want and then you get the program much faster than if you wrote it yourself in some other language.”

“OK,” said Jack. We continued on our project: getting a ChatGPT interface called AutoGen to program the Game of Life simulation for us. The Game of Life starts with a random pattern of cells on and off (see below).

The dots go on and off depending on the state of neighboring dots. Isolated cells and cells which are over-crowded die (disappear).  Healthy populations spawn new cells in their vicinity. The pattern keeps changing. Ok, maybe only something a nerd would love but it’s what we wanted to do.

“write code to display a game of life,” we told AutoGen. It did; a pattern flashed on our shared screen and then disappeared.

“write and save code to display a game of life,” we amended. AutoGen told us where it had saved the code, which I then executed manually. A static picture appeared on our screen rather than the animation which I had expected.

I looked at the generated code. “I think I see the problem,” I told Jack. “The animation is only set to run through five frames and then it stops. I’ll fix the code.”

“Why?” asked Jack.

“Because it’s broken.”

“Why are you fixing it?”

“Because…. Oh, I see what you mean.” I was fixing it because I’ve been a programmer for 71 (count’em) years and that’s what programmers do. Old habits die hard. What I had to do was tell AutoGen more precisely what we expected the code to do.

“write and save code to display a game of life. The animation should run forever.” The new code ran a long time.

However, after a while the screen entered a repeating pattern. If a state repeats once, the game is in a loop and gets boring.  If our screen were bigger, we might never see a repetition; but we only have the screen we have and didn’t want to be bored. “I know how I can fix the code so it doesn’t get stuck,” I said. Jack just looked at me until I realized that I had said English is the most powerful programming language.

“write and save code to display a game of life with an additional rule: every five frames, randomly either set one cell on or another cell off regardless of its current state or that of its neighbors.” It worked. I hadn’t written a single line of code. What I had done was act like a programming manager: 1) give a spec; 2) test the result; 3) refine the spec if it doesn’t result in what you expected; 4) repeat. Jack and I didn’t have to program but we did have to state specifically in English what we wanted.

Students don’t need to use slide rules, The Readers’ Guide to Periodical Literature, or a library catalog anymore. Nor do they need to spell obscure words correctly, do long multiplication, or have good handwriting. Typing will soon be obsolete. I don’t even use Google much. Why look at a bunch of ads (unless I’m shopping) when I just want an answer to a question? The ability to phrase a question or an instruction correctly, however, is more important than it’s ever been. So is the ability to look critically at what you get back and ask accurately and succinctly for sources and corrections.

The interfaces to AI are changing daily; we’re just in the early days. Much and probably soon more of our interface is spoken rather than written. The important skill is not in the use of ChatGPT or some other specific AI tool; what is essential is mastery of the language you use to access AI and in which it answers you.

Note 1: In case you want to see Game of Life in action, I asked AutoGen to make an HTML (web page) version. After a couple of tries, it got it right. Try it here.

Note 2. Maybe I should have just told AutoGen to find a way to stop any possible looping rather than designing the change myself.

Note 3: The rules for The Game of Life are (according to ChatGPT):

1. Birth: A cell that is dead at one step will be alive at the next step if exactly three of its neighboring cells were alive at the previous step.
2. Survival: A cell remains alive at the next step if two or three of its neighbors are alive at the previous step.
3. Death: A cell dies (or remains dead) if it has fewer than two live neighbors (due to underpopulation) or more than three live neighbors (due to overpopulation).

Note 4: https://github.com/tevslin/game_of_life is the github repository for the python code written by AutoGen for a Streamlit version of the modified Game of Life.

Note 5: AI can be accessed in many languages. However, it is currently best in English because the large language models have been trained on web material which is primarily in English.

See also:

Other AI posts in Fractals of Change.

The modified Game of Life

A Cease Fire in Gaza Will Not Make Hamas Go Away

It’s like ending crime by defunding the police.

I wrote to Vermont’s congressional delegation urging them to oppose Senator Bernie Sanders dangerous proposal to cut off aid to Israel ““unless there is a fundamental change in their military and political positions.”

Representative Becca Balint sent a long response (obviously a form letter but appropriate to what I asked). She writes:

“Hamas militants launched an unprecedented attack on 22 towns and army bases across Israel, killing at least 1,200 people including children and the elderly, wounding thousands more, and taking 199 hostages. This horrific terrorism is evil and I have joined my colleagues as an original cosponsor of a resolution condemning the terror attacks perpetrated by Hamas and reaffirming the U.S.-Israel alliance.”

Then she goes on to say:

“What is needed now is a true negotiated bilateral ceasefire. For the good of Palestinians and Israelis, a lasting bilateral cease-fire can only work if Hamas does not continue to rule in Gaza. Hamas is a terrorist organization, and its stated goal is to annihilate the state of Israel. It can’t remain in power in Gaza. It has violated international law by taking hostages and massacring civilians.”

She is right that a cease-fire can only work if “Hamas does not continue to rule in Gaza”. What she doesn’t say is whom Israel is supposed to negotiate the cease-fire with if not Hamas. Negotiating with Hamas perpetuates their rule. A cease-fire just prolongs the agony for both Israelis and Gazans. Israel must fight until Hamas leaders stop hiding among the civilian population, stop firing rockets at Israel, and surrender control.

Senator Peter Welch wrote in part:

“I fully support Israel’s right to pursue those who ordered and carried out the attacks of October 7^th. But Israel must do so in a way that does not lead to massive civilian casualties and the large-scale destruction of civilian infrastructure in Gaza. This will only incite more enemies against Israel and the U.S.  Israel’s pursuit of Hamas must be conducted in accordance with international law, which includes an obligation to protect civilian lives. Every effort must be made to prevent further civilian death and suffering. 

“That is why, on November 28^th, following the announcement of a temporary ceasefire between Israel and Hamas designed to facilitate the return of hostages and aid delivery to Palestinians trapped in Gaza, I called for an indefinite extension to that ceasefire.

“A ceasefire is critical to ensure that efforts to release the rest of the hostages are successful, to support the provision of humanitarian aid infrastructure in Gaza, and pave the way for meaningful negotiations for long-term peace and security for Israelis and Palestinians. In the intervening weeks, hostilities have resumed. I am deeply concerned about the immediate and devastating impacts on the civilian population in Gaza.“

The second paragraph contains the contradiction in his position. There was a cease-fire which led to the release of some hostages. Hamas stopped returning hostages; they resumed rocket-attacks on Israel; they repeated their promise to destroy Israel. But Welch apparently thinks the cease-fire should have been extended indefinitely anyway. How then is Israel supposed to “pursue those who ordered and carried out the attacks of October 7^th”? Sue them?

Senator Sanders didn’t reply.

Neither Balint nor Sanders, who are both Jewish as am I, are anti-Semitic. I’ve known Welch for decades and both like and respect him. Their positions are not Democratic-party orthodoxy; President Biden has been steadfast in rejecting a cease-fire which would leave Hamas in power. So why does our congressional delegation think Hamas can be conquered with a cease-fire or by denying military aid to Israel? Why do some people (often the same people calling for a cease-fire) think that crime can be prevented by defunding the police? History does not support the theory that tolerance is an effective way to stop evil behavior.

Hamas cannot win a war against Israel now that the US and Israel have dissuaded Hamas’ possible allies like Hezbollah and Iran from full participation in the conflict (for now). Hamas hangs on at great cost to the Gazans, among whom its leaders hide, in hopes that Israel will be forced into cease-fire. It’s a dangerous and cruel mistake to encourage Hamas in this hope.

See also:

Thank You, President Biden

Gaza Peace Depends on Hamas

How Not to Control Disease

Overstrict rules are counterproductive.

“All of the countries we’re visiting have a mandatory 10-day quarantine if you test positive for COVID,” said the cruise ship captain during the safety briefing just after we were instructed on how to don lifejackets. “The quarantine does not allow air travel.”

“COVID is just a bad cold.” he continued. “Remember, your captain told you that you will be quarantined for 10 days with no possibility of flying out if you test positive for COVID.”

Don’t test. Don’t test. Don’t test… people in the family groups whispered to each other. This was no MAGA crowd of vaccine-deniers; the vacationers on the US-flagged ship were mainly over-educated coasties like us as far as I could tell. As much as they were looking forward to seeing a different part of the world, they didn’t want to be stuck there, perhaps in an inferior hotel.

“You should take normal precautions against infectious disease, “the ship’s doctor said. “Wash your hands often. Use the hand sanitizers which are deployed around the ship. Cover up if you cough.” No mention of masks nor any visible supply of them. Nothing about staying in your cabin if you feel sick. And certainly no mention or sign of COVID test kits.

None of us had symptoms so I don’t know whether we would have insisted on testing if we did. Quite possibly not; we all had plans after the trip. Mary and I would have had a real dilemma (beside the ethics of possibly infecting other people) because we’re old enough so that COVID is not just a cold and we should get Paxlovid immediately if, despite our many rounds of vaccination, we are infected.

The point of this parable is that the 10-day quarantine rule, at least since we’ve gone from pandemic to endemic, is so strict that it’s counterproductive. It discourages people from testing and then taking reasonable steps if they’re infected with COVID or something else. Rules aimed at eliminating rather than just reducing risk are often riskier than the danger they seek to avoid.