FAQ/Q&A

First, many tech firms are announcing layoffs due to AI.

Microsoft cut ~6,000 employees in 2025, with software engineers disproportionately affected. Shortly before the layoffs, CEO Satya Nadella said AI was generating up to 30% of some Microsoft code.

Meta reduced workforce by about 10% globally while simultaneously shifting thousands of employees into AI-related roles.

In my opinion many firms cite AI when laying off because Wall Street analysts believe they are getting more efficient and the companies stock price rises.

Many of these layoffs are not due to AI but are being tied to AI because it drives up the stock price.

Second: 185 years ago in Canada, 80% of employment was in agriculture. Today it’s less than 2%. There were huge protests when farm mechanization – tractors and combines – came in. But we have survived farm automation.

Here’s the real story. Jobs change. New companies emerged. People had to learn new skill sets. Society as a whole changed. So, this is really a story about change.

Individuals and companies that don’t learn and apply these most powerful tools in human history will be left behind.

When the web came along in 1993 many people and firms couldn’t see the utility of it. But the web first, web native companies have seen the largest growth in the last 30 years. Firms that did not embrace it became less relevant, lost market share, went bankrupt or out of business.

So continual learning and committing to be an AI first employee and company is essential today. And in that vein, committing to be a lifelong learner and spending more time learning about AI tools will be essential for job security and organizational security going forward.

In a high stakes situation like a $10 billion merger between a Canadian and Japanese company, tens of thousands of dollars will be spent on professional translation to ensure the contract is agreeable to both parties.

For low stakes applications, people will choose to use the tools that I demonstrated. In other words, quality required is a function of context.

Methodology: when there is a billion-dollar merger there will be multiple translations. First the contract will be translated from English to Japanese. Then a separate translator will translate the Japanese contract back into English. The two English contracts will be compared, and discrepancies will be addressed in t he next translation back into Japanese. This will continue until there is a deep and consistent understanding by both parties.

In the same way, you an do a translation from English to German using one AI and then translate it back from German to English with another. In this way you can verify the accuracy of the translation and make corrections if necessary.

Wordly: Real Time Event Translation

I’ve been a professional speaker for 35 years. At major high-end conferences, real-time translation was once an expensive luxury. It required three to six interpreter booths, a team of translators, radio receivers, and headsets for attendees—easily costing more than $100,000 for a four-day, six-language event.

Today, organizers can use Wordly, an AI-driven platform that takes the conference’s audio feed and translates it into over 120 languages. No receivers needed — attendees use their smartphones and earbuds.

But it gets even better. Wordly also provides real-time transcription. I once sat next to a deaf French woman who was reading the speaker’s words — in French — on her phone as they were being spoken in English on stage.

So instead of offering just six languages, Wordly offers 120. Instead of costly equipment rentals, everything runs through personal devices. And instead of a $100,000 bill, it can cost as little as $1,500. That’s a 98.5% cost reduction—while also making events dramatically more accessible.

In other words, multinational organizations can drive cybersecurity awareness and training in multiple languages.

Consider how using AI tutors for cybersecurity at different levels might work. Individuals can ask AI to look at their individual systems and highlight weaknesses: teams can have AI look at cross-departmental work processes and workflows and identify missing links and fragilities; organizations can use “stacking” different AIs to take an organizational view.

AI is ushering in the greatest revolution that learning of every kind has ever seen. Using AI to beat AI will be the new norm, creating “good actors” working for security for our benefit will change the face of security.

In other words, many people think in polar opposites. Black or white. On or off. But there are many shades of grey.

In computer programming there is a concept called paired coding. If there are 10 people on a DevOps team week one, you’ll be paired with one person – and all week you’ll be learning informally from each other. Then in the second week you’ll be paired with another individual. This continues until you’ve been paired with everyone of the other nine on the team.

In other words, learning is continually embedded into the work process. Learning is informal and continuous.

The World Economic Forum’s (WEF) most recent job report said two things that are important to note: a) 30% of the average workers skill set is no longer relevant by 2030 and b) each employee needs an additional 20 days of learning and development per annum.

Finally, I believe the role of mentoring become more important for people with more tenure in an organization.

Stock market regulators like the SEC and the OSC require companies to report on materiality – in other words what matters to their bottom line.

A staggering 70% of all fresh water use globally is in agriculture. Water use for data centres is a tiny fraction of 1% of water consumption globally.

Sensors, AI, weather predictions, and smart irrigation can cut water used in agriculture by more than 20%. So, AI can save exponentially more water than running data centres to make the calculations and increase efficiency will use.

AI can also help municipalities with leak detection and increase the efficiency of industrial processes.

About 25–30% of treated drinking water worldwide is lost before it reaches customers. more than 126 billion cubic metres of treated water are lost globally each year through leaks and other distribution losses.

Again, AI can exponentially save more water than it consumes by being applied to this global problem.

Here’s my top dozen in order of daily use. I use:

ChatGPT ($20 a month) paid version 20-30 times a day.

Research, idea generation (brainstorming), image creation,

Gemini ($20 a month) paid version. Use is for verification, a second AI – like antagonistic prompting describe in the mergers and acquisition (MNA) translation example above.

Perplexity for research

Otter.ai for meeting transcriptions, summaries, action items. Competitors include Fireflies.ai,

Nano Banana for image creation

I used to use Sora for video creation

HeyGen for the video language translation that you saw (Synthesia is a competitor)

ElevenLabs for cloning my voice for

NotebookLM for creating podcast like discussion of written material.

I also use free versions of Claude, Grok, (these are LLMs like ChatGPT and Gemini)

Just for fun

Suno for creating songs and music

Claude Code for building apps

Many of my clients use Copilot because they are a Microsoft shop. If this is the case in your organization become a ninja using it.

Here’s an anecdote from my new book released in February 2026 on Amazon.ca entitled The Age of Fakes!: How AI Abuse, Fake News, and Deepfakes Threaten Business and Society

A Cautionary Tale

I was on a TV show talking about artificial intelligence (AI) recently. The host, who is in the public eye, shared an experience. There are hours of video of her on the TV station’s site and YouTube. Scammers had trained an AI on her voice. Her family received a phone call that sounded EXACTLY like her that went something like this:

Her family has a safe word. If a family member doesn’t use it in this kind of emergency phone call, everyone knows that it’s not them. For this example, I’ll make one up: “jalapeno peppers.” If it’d really been her, she might have said, “I’m in the hospital and am starving and want to have a pizza with “jalapeno peppers.” Her family then would have known that it was really her.

So, the question is: Have you done safe word training with your family? Grandparents are particularly vulnerable to this kind of fraud.

https://www.amazon.ca/Age-Fakes-Deepfakes-Threaten-Business/dp/3982801001/ref=sr_1_1

I am a very public figure. There’s lots of video and audio of me up TV and radio stations, podcasts, YouTube and social media. Anyone can use the tools described above to clone my voice using ElevenLabs of Synthesia. So, I have to focus on mitigating risk, and cybersecurity. So, we conduct cyber security training with everyone at my organization. We have a password manager and use MFA (multi factor authentication), we patch and update all software regularly and we

For this question I’m going to answer with a methodology. Whenever there’s something that I don’t understand, I go to ChatGPT and say explain this in a way that a 7-year-old can understand.

So, here’s my prompt: Describe a token used by an LLM in such a way that a seven-year-old can understand.

Here’s the reply: Imagine you're building with LEGO.

A token is like one LEGO block.

When you ask AI a question, it doesn't see whole sentences the way people do. It breaks your words into lots of little pieces—its LEGO blocks.

The AI puts those blocks together to understand what you said and build an answer.

A token is a tiny piece of a word that AI uses like a LEGO block to build understanding and answers.

So, tokens are the building blocks of AI thinking.

An important part of prompting is iteration. So, with this I would ask multiple questions. Like

Why are tokens important?

How many tokens are used in an average prompt.

How many tokens are typically allocated to answer queries in the free version of ChatGPT.

In the free version of ChatGPT, the maximum response length (output) is capped at 4,096 tokens per single answer. This translates to roughly 3,000 words of generated text. If a response requires more than this amount, the model will cut off mid-sentence, though you can prompt it to "continue" to generate more.

How many tokens are typically allocated to answer queries in the $20 a month version of ChatGPT.

In the $20/month ChatGPT Plus tier, OpenAI does not allocate a strict per-query token maximum. Instead, it provides a large "context window" that holds your entire conversation history, paired with usage caps based on message volume rather than exact token counts.

Maximum Response (Output): While the context is massive, a single generated response is generally capped dynamically (often between 4,000 and 16,384 tokens, depending on the specific model).

How many tokens are typically allocated to answer queries in the $200 a month version of Claude Code?

Anthropic doesn't allocate a fixed, static number of tokens for queries. Instead, the $200/month Claude Code plan (known as the Max 20x plan) provides 20 times the usage allowance of the standard Pro plan.

This translates to roughly 220,000 tokens per rolling 5-hour window. Depending on the complexity of your prompts and how much conversation history the tool is processing, this typically yields 100 to over 800 prompts per session block.

AI is going to be embedded in job, discipline and profession over time. So, I would ask my kids what are they passionate about? And then have them use AI in the pursuit of their passion.

Joseph Campbell, the American mythologist and author, popularized the phrase "follow your bliss." Campbell defined bliss as your deepest sense of aliveness, joy, and authentic purpose. He believed that by aligning your actions with this internal calling, allowing doors to open that wouldn't open for anyone else.

ChatGPT or Claude is a great place to start. Begin with just the free version. These are great all-purpose Ais that you can use. Here’s a trick: use AI to learn how to use AI.

If you want to have fun creating images, use Nano Banana (the Gemini AI image generator) is available through for free or paid versions. Here’s a headshot that I generated with it. (A friend said did you use the Brad Pitt good looking filter in your prompt?)

Spend time playing with and learning about AI tools. There are now more than 30,000 tools for business.

Jensen Huang was on the Le Fridman podcast in March 2026. When asked about job loss he talked about the profession of radiology.

AI had superhuman capabilities in the field in radiology in 2019 or 2020. Experts predicted AI would wipe out the profession. While AI did take over the task of reading scans, it allowed radiologists to study more images faster and diagnose diseases more effectively. Because this increased their productivity and throughput, the demand for human radiologists actually grew.

Huang dismissed the notion that AI is making software engineers obsolete, noting that NVIDIA’s engineers use AI to code, yet the company is hiring more people than ever. This is because the core purpose of a software engineer is high-level problem-solving, teamwork, and innovation, not writing lines of code. People need to a distinction between tasks and purpose. Are there more problems for engineers to solve? Can they solve them faster using these tools?

In the same way actuaries used to have a very limited number of models to predict longevity. Before the 1990s, many pension plans systematically underestimated improvements in life expectancy. I predicted that we will see an explosion in the number of models used on a workforce staff to predict. So instead of a thousand model for one company, using these powerful tools that may grow 10-fold or 100-fold. So back to Jensen’s paradox we may find there is need for MORE actuaries, with far better skills to accurately predict the liabilities of a book of business before buying it.

Watch a 13-minute version of this at https://www.youtube.com/watch?v=N_wq9OV4T4c

The only way we learn is by doing. So, start playing with the tools. Have fun. Most large organizations in Canada invest in training and development. Make sure you have met with your manager and mapped out your learning path. Here’s the prompt I would start with:

I am an actuary for one of Canada’s largest firms. what questions should I ask when I meet with my manager to ensure 1) I am able to add most value in the organization; 2) I am going to increase my skillset which will increase my employability?

People only learn by doing. I’ve never been able to read my way into physical fitness. Make a commitment to play with AI for 15 minutes every day. Take courses. Talk with others who are using apps and discover what AI use cases they find most beneficial and interesting – both personally and professionally.

My wife and I are medical ninjas in helping our family and friends with medical problems that they are facing – a use case that you might not think of. Here are a couple of resources:

Twelve Techniques for Prompting: https://www.dropbox.com/scl/fi/9h76wmj85pucperw1nwhr/Twelve-Techniques-for-Prompting-ver-2.0.docx?rlkey=rvo0gt5y8oxo4zb3rzkyi252a&st=6ldgok98&dl=0

FREE AI prompting course: https://www.sectionschool.com/events/live-events/introduction-to-ai

Let me know what insights you get from this by emailing me at jim@jimharris.com

Yes & no. It depends on how you look at the question and the answers that flow from it.

I had the pleasure of being a journalist in an interview with Demis Hassabis in 2024 before he won the Nobel Prize for AI.

It used to take a PhD student an entire 5-year career to exactly map out the 3D structure of a single protein. Google Deep Mind’s AlphaFold was able to do that work in 2 seconds. So AlphaFold mapped 200 million proteins in a single year – all human and all animal proteins.

Why is this important? Proteins are the building blocks of life and how our bodies organs communicate with each other. Would it be important to know which part of a protein goes rogue and creates cancer at stage zero? Yes? No? Maybe?

When you do the math on this, AlphaFold was able to accelerate learning 1,000,000,000-fold. This billion X improvement – from 5 years to two second means that humanity gets a billion years of insight into disease. And now researchers can focus on curing cancer at stage zero. Demis is talking about potentially putting an end to major disease categories within a decade.

So, in some ways the 100X leaning improvements is drastically under rating the result.

From another perspective, learning is an organic process. A 2-year-old can understand advanced trigonometry no matter how many AI tools they have access to.

But once an individual has mastered certain critical thinking skills, AI can dramatically accelerate their learning.

I am a writer and author, and using AI is like having a PhD level research assistant at my beck and call 24-365 for $20 USD a month.

So, this is a hard question to answer and can be answered many different ways.

The advanced voice feature of ChatGPT allows you to have a conversation about any topic. I know of a CEO who when driving from Vancouver to Whistler has a conversation with ChatGPT about any topic he’s interested in. It’s like having a personalized podcast focused on the topics you are most interested in at the moment.

When I was in high school, calculators were just coming out. Educators and parents feared that no one would be able to do math if they relied on these tools.

High schools, universities and the teaching profession is going to have to change how it teaches. Some colleges and universities refuse to let students use AI. For me that’s akin to saying don’t use the telephone in 1900. Or don’t use the internet in 1993. This type of philosophy only handicaps students for the future.

What educational institutions need to do is focus on analytical thinking, critical thinking, and creative thinking. Here’s the insights from the World Economic Forum (WEF) Future of Jobs Report 2025:

Some teachers and university professors are changing the way they mark students – with up to 80% of the class mark given for participation and demonstration of critical thinking in class discussions.

Some students will always work to take the path of least resistance. In my time in university some students hired other to write their essays. Others used Coles notes. So, they used any means possible to avoid developing critical thinking skills.

The insight for professionals in the workforce is every individual needs to take responsibility for their learning path.

My second book The Learning Paradox published in 2002 argued that 80% of the technology that we will use in our daily lives in just 10 years hasn’t been invented yet. So, job security is based on learning, changing and thriving amid uncertainty. And what adults fear the most is learning, changing and uncertainty. So, our very job security as individuals and organizational security for corporations is based on continually learning.

Any thing that is dull and repetitive will be automated over time. So, typing number from one excel spreadsheet into another will disappear.

The skills discussed in question #14 will become more important. Proficiency with AI will be essential.

In our company we hire two coop or student interns a year. The most recent interview process had 66 students applying for the position. One of the screens was heavy use of AI. If they didn’t use AI, they had no chance of making it to the short list. We interviewed 9 students and made an offer to only one, who happened to be the heaviest user of AI relative to his peers. There were other criteria such as English language proficiency, ability to listen and respond, and whether they would work will in a team.

See answers #8, #11 and #14

Try this prompt and see what answers you get and see if you agree: What should students in high school study to prepare them for the age of AI and the workforce?

Here’s the answer I got from ChatGPT – and I think its pretty good:

Most important subjects to study

Skills they should deliberately build

1. Learn how to ask better questions
2. Use AI as a thinking partner, not a cheating machine
3. Learn basic coding: Python is ideal
4. Get strong at spreadsheets and data
5. Practice public speaking
6. Learn to write clearly
7. Build things: apps, videos, research projects, small businesses
8. Understand bias, privacy, copyright, and misinformation
9. Develop teamwork and leadership
10. Get comfortable learning new tools quickly

Best practical advice

Students should graduate high school able to say:

“I can use AI to research, write, analyze data, build simple tools, explain my thinking, and solve real problems.”

That combination — human judgment + AI capability — will matter more than memorizing facts.

What should students in university or college study to prepare them for the age of AI and the workforce?

I leave it to you to get the answer yourself from ChatGPT, Claude, Perplexity or your favourite LLM.

Realistically that won’t happen. Most people have no idea what AI can do. In the graph below, each dot represents 3.2 million people. (2,500 dots = 8.1 billion people). The colour shows the type of AI interaction that person have had.

Grey = NEVER used AI

Green = 1.3 billion free chatbot users

Yellow tiny sliver = 15-25 million who pay for it

Single red dot = 2-5 million people using AI for coding and development.

Most people on earth have absolutely no idea what AI can do right now.

Here’s a different way to look at the question:

ChatGPT hit 100 Million Users in Just 2 months Making it the fastest adopted new technology in human history

1. The Telephone (1876): It took 70 years to reach half of US households. The barrier was massive: telcos had to physically string copper wire to every front door.
2. Radio (1920): Radio hit 46% adoption in 10 years.
3. Television (1946):  Television rocketed from 1% penetration to 75% by the early 1950s. But the cost of adoption was severe. In 1948, a television set could cost $400 – equivalent to $5,000 today.
4. The Internet (1990): It took roughly 10 years to reach 43% of Americans. Why? It required an expensive computer ($2,000+), a monthly subscription, and a learning curve.
5. Generative AI (2022): The Zero-Cost Revolution If the telephone was a slow climb and the internet was a steady hill; AI is a vertical launch.

EY has invested $1 billion in AI for its 400,000 employees. See my video on this at https://www.youtube.com/shorts/MWUw_s6JnXE

An example of a low ROI is where a company gets Microsoft Copilot for every employee at $25-$40 a month – but only 20% of employees are using it.

A key insight is that organizations have to invest in

1) training;

2) providing access to paid tool to people who will actually use them;

3) socialize high ROI use cases so that employees understand how it can benefit them; 4) Use the paired programing or federated learning model discussed in answer #3;

5) align systems and structures. One organization I know would not let any divisional President attend the global annual leadership summit unless they had completed certain basic AI course. Similarly, when employee’s annual evaluation has a component tied to taking and passing certain AI courses the training has traction. Just like if you are in a profession like being an actuary you have to have a certain number of hours of CEU (continuing education units) completed to maintain your professional designation.

HeyGen see https://bit.ly/3LylxAk

See answer to #17

When AI can create 1-billion-fold increases in productivity (Google DeepMind AlphaFold example) and potentially eliminate major classes of diseases in the next decade, it is not a fad.

AI can be empathetic already:

Some organizations and startups are testing AI mental-health companions, including:

1. AI therapy bots
2. School-support bots
3. Mental-health coaching systems

These are generally positioned as:

1. companions,
2. screening tools,
3. or support tools,

Every new technology is used for both good and bad purposes. The bad actors are in fact the fastest to adopt new technologies.

Read the lead chapter of my new book of which I am one of eight coauthors:

The Age of Fakes!: How AI Abuse, Fake News, and Deepfakes Threaten Business and Society

https://www.amazon.ca/Age-Fakes-Deepfakes-Threaten-Business/dp/3982801001/ref=sr_1_1

My chapter is the lead chapter and Amazon.ca gives it away for free as the sample. On the page search for the Read Sample button under the book cover image and you can read it for free. The essential take aways are:

1. Invest in mandatory cybersecurity training for every employee.
2. Ensure onsite and offsite automatic backup procedures to protect your data.
3. Implement multi-factor authentication (MFA).
4. Have your training firm run simulated phishing attacks.
5. Hire an outside firm to perform vulnerability testing.
6. Make sure your IT team regularly updates the software with security patches.
7. Gamify it: reward employees who pass the test with a free firm-branded coffee mug.
8. Have different employees lead the training every quarter. The best way to learn something is to teach it. Stephen Covey, who wrote The 7 Habits of Highly Successful People, used to always promote learners becoming teachers.
9. Get cybersecurity insurance for your company. Our insurance firm gave us lower rates once they conducted penetration testing and cyber readiness testing for our systems.
10. Include cybersecurity training as a mandatory part of onboarding before granting new employees access to IT systems.

Banks insurers need to actively work to accelerate the speed of innovation, otherwise disruption will ensue.

Nubank began in Brazil in 2013 as a start up. In 2026 it has more than 135 million customers. Reuters has described it as "one of the fastest-growing digital lenders in the world." It has posted exceptional profitability for a growth-stage bank, with return on equity reaching about 33%, higher than many traditional banks.

In 2007 75% of Kenyans were unbanked. The mobile telco company Safaricom launched M-Pesa ("M" for mobile, "Pesa" is Swahili for money) in March 2007. It turned the 20,000 cell phone kiosks in the country into banks in essence.

In China Alipay was begun in 2004 and now serves 1.3 billion merchants and customers (mostly in China). Payment volumes are estimated at $21 trillion USD a year.

All three examples are digital first firms that have dominated their respective markets first then spread out globally. A digital first firm will be faster to implement AI than a traditional firm.

These are three examples of how disruption occurs from the bottom up. Underserved or not served at all markets provide opportunities for new entrants to develop products and services, gain traction, increase revenue and expand up market.

A friend and colleague Salim Ismail has written Exponential Organizations 2.0. (I am quoted on page 47). The book outlines 11 principles to accelerate that rate of innovation in an organization. You can buy it for $20 Canadian in paperback or $8.33 for Kindle or get ChatGPT to write you a summary of the 11 key principles for free.

https://www.amazon.ca/Exponential-Organizations-2-0-Playbook-Growth/dp/B0CJ4F49CH/ref=sr_1_1

If you wants to implement this framework I can help.

If I prompt ChatGPT: Tell me everything that there is to know about Queens, I get a terrible answer.

Why? Because I provided no context.

Was I talking about Queen’s university where I graduated from?

Was I asking about Queens in New York City next to the Bronx?

Was I talking about how a hive organizes around the Queen bee?

Was I asking about how a queen moves on the 64 squares of a chess board?

Was I inquiring about Freddie Mercury the front man for the band Queen?

Was I referring to a drag Queen?

Or the recently deceased Queen of England?

I got a terrible answer because I provided no context. And then I blamed the tool.

A master magician can accomplish impossible things with a magic wand.

But the same wand in the hand of the sorcerer’s apprentice results in terrible outcomes.

And the result has nothing to do with the wand. And everything to do with the skill of the one wielding it. That’s why education is so critical in improving performance of using AI. I urge you to take the following FREE online AI prompting course: https://www.sectionschool.com/events/live-events/introduction-to-ai

Taking the above prompting course dramatically improves the quality of answers that you will receive.

Large Language Models (LLMs) can hallucinate–in other words, make stuff up. The good news is that with each successive release of LLMs hallucinations have declined dramatically. Nonetheless, what can you do to reduce hallucination?

You can specify “cite only credible sources such as The Wall Street Journal, the Globe and Mail, Wired magazine etc. Define what you mean by credible. Also require the LLM to provide URLs” Then read the source documents.

For very important queries use antagonistic prompting

Antagonistic Prompting: This is something that you heard here first. Take the output from your query from ChatGPT and put it into Gemini (Google) and prompt as follows: “Here is the answer from ChatGPT for my query X. [insert query]. What are the deficiencies in the answer? Please refine the answer.

Take the refined answer from Gemini and put it into Anthropic’s Claude and do the same again. Take the answer from Claude and put it into Perplexity and do the same.

Each LLM has strengths and weaknesses. Try this ChatGPT query: “What are the strengths of ChatGPT, Gemini, Claude and Perplexity relative to each other?

It depends on the human. I use AI as a thought partner. As a researcher.

If I want to write an article of how the use of APIs are driving value for business I will prompt ChatGPT

What are the top 20 ways that APIs are driving radical new value for businesses in X field. Cite credible sources such as (fill in) and provide URLs.

Then I will look at the list and say 1, 7 and 19 are really interesting to me. Write 500 words on each with credible sources.

Then what would the harshest critics say about those three 500 words pieces? Write a 500-word critic of each with at least three points in each critique. So, I am getting the LLM to debate two sides of a question.

Then who are the world leading experts on both sides of the debate. What is their contact information? And their definitive article on each of these six 500-word pieces.

I then start with all this and write my article. So, AI didn’t write the final article. It was a PhD level research assistant to me.

No

But my headshot in answer #9 was created by AI with a very elaborate prompt.

This question is too hard to answer without more context.

May SaaS software providers are embedding AI tools within their product. They have to, otherwise they will lose customers to competitors.

I like to use the market share leader in any AI category. For instance, I use Otter.ai for transcription. Why? Because it is transcribing a million words every minute globally.

My theory is that it will have the most revenue to reinvest in improving.

I begin by using the free version to test it. Then after that runs out or I start hitting use limits after I’ve determined it’s delivering real value, then I pay for a month-to-month subscription. Only after 3 months of consistent use with proven ROI to I subscribe to an annual plan. (usually offers a 20% saving but you pay for a whole year at once). The I share the login with my spouse. So, a $20 USD subscription drops to $10. Finally, if you are a student or teach at a university you can get discounts.

Use an LLM to find the

Was this Yoda asking?

Some organizations are overpromising and hyping their AI elements – because they fear bleeding customers to startup competitors. So yes, there is hype in the market.

But when you can get 1-billion-fold productivity gains (Google DeepMind AlphaFold example in question #12) the transformative power of AI is actually significantly underhyped.

We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.

Productivity – meeting transcription, proposals – Otter.ai

But really, I need more context to properly answer this question.

Language translation: HeyGen see https://bit.ly/3LylxAk

In some cases, but probably not in the way you might be thinking. See answer #23.

When I was writing about Nubank. M-Pesa and Alipay, I have written about them in books, articles and blogs. And talked about them in videos. But in some cases what I’ve published is years old. So, I know the story, the background and I use AI to instantly update me – for instance the last time I wrote about Nubank was two years ago and it only had 100 million customers. Now it has 35% more.

I do believe there will be a rush to build more data centers when we are not using existing data centers efficiently. Read my article https://corporateknights.com/energy/the-single-greatest-tool-for-reducing-greenhouse-gas-emissions-is-energy-efficiency/

Amory Lovins one of the world’s leading energy efficiency expert notes that there is 200GW of available data center processing if existing data centres were used as efficiently as possible.

Also see answer #1. Some CEOs are blaming AI for layoffs because it results in their stock price going up. Some argue that they would have laid off employees even if there was no AI. Economic uncertainty, tariffs, the US & Israeli war with Iran are all reasons that CEOs are nervous about the future economy and laying off people. But citing those reasons doesn’t drive up stock price. Citing AI does.

I don’t know if anyone is going to read all this, but if you have any questions, feel free to email me at jim@jimharris.com