Welcome to Issue 3. This time: ChatGPT gains a new way to answer with charts and calculators you can adjust, a $1.8 billion partnership aims to use AI for medical research, AI agents head for PCs and the workplace, and AI tools turn up in a run of bank hacks in South Korea.

Plus: a Deep Dive on restoring old family photos with AI, three reading picks, a look at investment schemes that borrow the “AI trading bot” label, and a 10-minute exercise: a calculator with a slider and a graph.

AI SIG announcements

“AI, Medicine and You” at the General Meeting, Thursday, October 22

Chris Brigham, MD, will present “AI, Medicine and You” at the Computer Club’s General Meeting on Thursday, October 22, at 2:30 pm in Pinckney Hall. A recording will be available for those who can’t attend.

Welcome, new members!

The AI SIG now has 434 members, and the number is still growing. Welcome, and thank you! If you’re new, past issues of this newsletter are in the archive.

AI classes coming this fall

With the Computer Club’s Education Program back in full swing after the summer, I’m preparing a few AI classes. The first is “ChatGPT: The Parts You’re Not Using (Yet).”

Most people know ChatGPT as a tool for asking questions, but that’s only scratching the surface. The class explores capabilities you may not have discovered yet, from voice conversations, research, image creation, and personal health and finances to building websites and having AI agents perform tasks on your behalf. Through practical examples and live demonstrations, we’ll tour ChatGPT’s growing collection of tools, including Projects, Plugins, Work, Sites, and more. We’ll also compare what’s available on computers and smartphones, explain the differences between free and paid plans, and discuss how to use ChatGPT safely and effectively. No prior ChatGPT experience or technical background is required.

Attendance is limited to 15 and sign-up is required. Watch for the official announcement in the Computer Club’s Education Weekly newsletter for the date, time and sign-up details. What else would you like to learn? Send class requests to SCHHCC.AI.SIG@gmail.com.

SIG meetings

We’re still working out rooms and times for regular AI SIG meetings, and it has turned out to be more involved than expected. The timing didn’t help, since we started during the recent administrative disruption that affected many association services. It may be time for Lifestyles to consider adding an AI agent to the staff. In the meantime, the humans are still on the case, and we’ll share the details in an upcoming issue of this newsletter and in Computer Club mailings.

Have a question or an idea for a future AI SIG session? Send it to SCHHCC.AI.SIG@gmail.com.

AI news worth knowing

ChatGPT now answers with charts and calculators you can adjust

On Wednesday, October 7, OpenAI released a ChatGPT feature called Intelligent UI, along with GPT-6, the newest AI engine behind ChatGPT (see our guide to AI model names). In plain terms, ChatGPT can now answer with more than words: buttons, calculators and charts you can change, diagrams, even a map for a hiking trip. TechCrunch’s examples include a diagram of how a wing generates lift and a personal savings calculator. You can dial the visuals back if you prefer plain answers.

Plus, Pro, Business and Enterprise plans began getting it that day, and Intelligent UI has since reached Free plans as well. It isn’t clear from the reports I could read whether free accounts get GPT-6 itself or only the new display.

Why it matters: Many of us learn best by seeing and trying. A savings calculator you can adjust, or a diagram of how something works, can explain more than a paragraph. A polished display doesn’t make an answer correct, so check any numbers that matter, especially about money or health. The newest models often start on paid plans, so free users may not always get the latest engine behind the new display.

Want to see it for yourself? Try it below walks you through asking ChatGPT for a savings calculator with a slider and a graph, much like the one in TechCrunch’s examples, and then checking its math.

Sources: ChatGPT is getting a lot more visual with the launch of a new interface, TechCrunch, Lucas Ropek, Oct. 7, 2026 · GPT-6 and Intelligent UI for everyone, OpenAI, Oct. 7, 2026

A $1.8 billion push to build biology data for AI

On Wednesday, October 7, Biohub, the research organization funded by Mark Zuckerberg and Priscilla Chan, announced a partnership with the U.S. Department of Energy, the National Institutes of Health, Google DeepMind, Meta and Isomorphic Labs. They plan to measure how living cells respond to many kinds of change, then use that data to train AI that predicts how cells behave, a “virtual cell.” Scientists could then test some ideas on a computer first, which may speed up drug research.

The $1.8 billion adds several pledges together, including datasets the NIH already paid for, so it isn’t all new money. The data is to be made public eventually, but commercial funders get an early period of exclusive access, one year according to The Next Web. Biohub expects a first dataset in about a year and accurate predictive models within five years. Those are goals, not guarantees.

Why it matters: Developing a new medicine takes years and a great deal of money, and this is an attempt to use AI to shorten that road. It is early work and no one is promising cures, but it shows that some of AI’s most meaningful uses may be in medicine rather than chatbots. If that interests you, Chris Brigham, MD, presents “AI, Medicine and You” at the Computer Club’s General Meeting on Thursday, October 22, at 2:30 pm in Pinckney Hall (see Announcements).

Sources: US government, Google join Zuckerberg-backed Biohub in $1.8 billion push for AI biology data, Reuters via The Star, Krystal Hu, Oct. 7, 2026 · Biohub, Meta, Google DeepMind and the US pool $1.8bn for AI biology data, The Next Web, Oct. 7, 2026

AI agents head for your PC and the workplace

Two big companies showed where AI assistants are heading: from answering questions to doing things for you on your own computer. These are called AI agents, programs that carry out tasks on their own rather than just answering questions. On Wednesday, October 7, Microsoft and Nvidia introduced the Surface Laptop Ultra and similar laptops from other makers, designed to run AI on the machine itself rather than only on distant company computers. Pre-orders start at $2,599, and partner laptops ship October 16. Microsoft also made a safety feature, Microsoft Execution Containers, generally available. It lets developers and IT staff fence an AI agent in to only the files, network locations and tools it has been given.

The next day, Google Cloud announced a “universal agent for work,” an assistant for its business customers that plans a task, uses tools and hands back finished work. It is in private preview, and I found no plans or prices.

Why it matters: None of this is aimed at home users yet, but it is the direction of travel, and it raises a question we already know how to answer with people: how much access do you hand over? You wouldn’t give a new helper the keys to every room. When an AI assistant arrives on your own devices, look at what it asks to open, such as your email, photos or files, and say yes only to what the job needs.

Sources: Windows, Surface And RTX Spark Event: Everything AI Microsoft Unveiled, HotHardware, Oct. 7, 2026 · Gemini at Work, Google Cloud, Oct. 8, 2026 · Google Launches Gemini Agent for Enterprise Work, Let’s Data Science, Oct. 8, 2026

AI tools used in hacks of South Korean banks

On Wednesday, October 7, the security company CrowdStrike reported that a hacker used AI tools to break into South Korean banks and other financial firms from late September into early October. Reuters counts at least nine banks targeted, and South Korean police opened an investigation on October 6. The hacker, probably a Chinese speaker working for profit, combined a free security-testing program with Anthropic’s Claude Code, a coding assistant, and other AI models. Shinhan Bank says personal data on about 25,000 customers was exposed; KB Kookmin Bank put its figure at 119.

CrowdStrike learned this from records the hacker left in open folders on a server. It has not tied the activity to any known group, and the suspect’s reported age and location are unconfirmed. These reports concern South Korean banks.

Why it matters: Most of us bank online or by phone, and criminals are borrowing the same AI tools we’re learning about. AI didn’t act on its own here; a person used it. The best defenses haven’t changed: turn on your bank’s account alerts, use two-step sign-in, and if a call or text claims to be your bank, hang up and call the number on your card.

Sources: CrowdStrike says China-based suspect used AI tools in South Korean bank hacks, Reuters via iTnews, Oct. 8, 2026 · CrowdStrike links South Korean bank breaches to AI tools and a China suspect, The Next Web, Oct. 8, 2026

Deep Dive: Restoring Family Photos with AI: How Much Is Real?

A faded sepia snapshot of three adults, two babies and a young girl standing on a lawn.

Original

The same photo restored in muted color with a pale blue sky. The people and the scene match the original.

Restored by GPT Image 2

Before and after: the faded 1930s print and the restoration by GPT Image 2, the one that stayed closest to the original in this test. Click either picture to read the full article.

Old family prints are easy to love and hard to look at: faded, scratched, faces gone soft. AI image models, the engines behind tools that create and edit pictures from written instructions, promise to repair them and even add color. This Deep Dive repeats a spring experiment with two newer releases, OpenAI’s GPT Image 2.5 and Google’s Nano Banana 2.1, running one faded 1930s family snapshot through five models with the same instructions.

Key takeaways:

  • Repairs come with inventions. The models sometimes added things the photo never showed, such as a dirt road, a fence post, or a car, and some smoothed skin or changed a face.
  • The colors are guesses. The same dress came out cream, blue, white, or yellow depending on the model and the run.
  • Software can’t vouch for the result. An automated face-matching check passed nearly every changed face, so you still have to look yourself.

It’s one print and one person’s judgment, not a ranking of the models, and it closes with practical tips for trying this at home. It will interest anyone with a shoebox of old photos. The full article has side-by-side comparisons.

Read “Restoring Family Photos with AI: How Much Is Real?”

Worth a look

  • AI Slop Is Dangerous: Here’s How You Can Detect It (AARP, updated March 2026). Background reading on the flood of low-quality, AI-generated posts, pictures and videos in social media feeds, with practical tips: check who posted it, don’t share while you’re upset, and use a free tool to check an image’s origin. Free to read.
  • AI Customer Service Scam Study (Better Business Bureau, 2026). How scammers pose as customer support for AI products and use fake “sponsored” search ads and lookalike websites. Reports to the BBB’s Scam Tracker that mention AI rose from 782 in 2023 to about 3,000 in 2025. Free; there’s also a shorter question-and-answer version.
  • The State of AI Report 2026 (Nathan Benaich). A wide-ranging look at AI’s progress, from useful new tools and medical research to growing cybersecurity risks. Written from an AI investor’s perspective, it’s a well-sourced overview of what’s happening and what may come next, with plenty of links for those who want to dig deeper.

Stay safe: Investment schemes dressed up as AI trading bots

A good reminder, prompted by recent SEC cases.

On September 29, the SEC announced lawsuits against four companies it says used the promise of “AI trading bots” to take more than $15 million from investors. According to news reports, the SEC alleges that one pair of companies said it would use clients’ money to “rent” AI trading bots that did not exist, while another showed investors fake profits on a website, then told those who tried to withdraw that their accounts were “frozen” until they paid more. These are allegations; the cases haven’t been decided.

Know the signs: promises of steady or guaranteed returns, “AI” offered as the secret ingredient, pressure to act fast, claims of government approval you can’t confirm, and fees demanded before you can withdraw your own money.

  • Before investing, look up the person and firm for free at Investor.gov. A fancy website proves nothing.
  • Never pay a fee to get your own money back.
  • If you’ve been targeted, contact your bank right away and report it at ReportFraud.ftc.gov.

Sources: Fake SEC Credentials and AI Trading Bots Allegedly Tricked Investors Out of $15M, IBTimes, Sept. 2026 · Fake signals, frozen accounts: SEC sues crypto AI platform over $12.5M, InvestmentNews · Investor.gov, U.S. Securities and Exchange Commission

Try it: Build a savings calculator with a slider and a graph

You’ll learn how to ask ChatGPT for a small interactive tool, the kind of thing its new Intelligent UI feature (see the first story above) is built for, and how to check that its numbers are right.

Time: 10–15 minutes. Tool: ChatGPT, free to use, and signing in is optional. Use only the made-up numbers below. Intelligent UI is now available on the free version, but what appears can vary, so there is a fallback below.

  1. Go to chatgpt.com and start a new chat. You can sign in if you have an account, but you don’t have to.
  2. Paste the request below and press Enter.
  3. Look for a slider and a graph. Move the slider and watch the graph and the numbers change.
  4. Check the answer with a calculator or your phone: at 3%, the 10-year balance should be $6,719.58; at 4.5%, $7,764.85. (On a calculator: 5,000 × 1.03^10.)
  5. Paste the follow-up request below.

The request:

Build me an interactive savings calculator in this chat, if that feature is available. Use these made-up numbers: a $5,000 starting amount, interest added once a year (compounded annually), no extra deposits, over 10 years.

Include:

  1. A slider for the yearly interest rate that goes from 2.5% to 4.5% in steps of 0.25%, starting at 3%.
  2. A line graph showing the balance at the end of each year, from year 0 to year 10. The graph should redraw when I move the slider.
  3. Large, clearly labeled numbers for the ending balance and for the total interest earned.

Add a one-sentence plain-English explanation of how the balance is figured. If you can’t make an interactive tool, give me a table with the 10-year ending balance for each rate from 2.5% to 4.5% in 0.25% steps instead. Don’t use outside services, and don’t give financial advice.

The follow-up:

Add a second line to the graph showing what the balance would be if the interest were not compounded, just $5,000 plus the same dollar amount of interest each year. Then tell me in one sentence how far apart the two lines are at year 10 at a 3.5% rate.

You’ll know it worked if the slider moves in quarter-point steps, the graph redraws as you move it, and the numbers match the check above. At 3.5% the follow-up’s two lines should end $1,177.99 apart ($7,052.99 compounded, against $5,875.00 with simple interest). The curve bending upward is what compounding looks like.

If something is off, ask ChatGPT to “show your arithmetic step by step” and compare. If the graph doesn’t change when you move the slider, say so and ask it to fix that. A polished display isn’t proof the math is right, so checking it yourself is the skill to keep.

If no slider appears, the request asks for a table of the 10-year balance at each rate instead, and it will have the same numbers. These are illustrations with made-up numbers, not financial advice, and you should never type real account numbers into a chat.

If you’ve used AI for a practical job, we’d love to hear about it: email SCHHCC.AI.SIG@gmail.com with a subject starting “Share:” and tell us how you’d like to be credited (full name, first name and last initial, or anonymous).

That’s it for Issue 3. Tell me what helped, what confused you, and what you’d like covered next at SCHHCC.AI.SIG@gmail.com.

Until next time,
Ron and Jim

Prepared with AI assistance.