|
Today's Tip: The 2027 Social Security Raise Nobody Is Doing the Math On Everyone is talking about a 3.8% Social Security raise next year. Almost nobody is talking about what's going to eat most of it before it ever hits their checking account. Let me explain… My friend Dot Kowalski is 67. Retired nurse, 41 years on the floor at a hospital in Erie, Pennsylvania. The woman has run a household budget on a spreadsheet since 1987. She is not someone you sneak a price increase past. A couple of weeks ago Dot called me, excited. She had just read that the 2027 Social Security cost-of-living adjustment — the COLA — is forecast to come in around 3.8%. That's up from 2.8% this year. The Senior Citizens League is calling it one of the biggest bumps in years. Dot pulled up her calculator. Her current check is $2,140 a month. A 3.8% raise meant about $81 more a month starting in January. She was already mentally spending it… new tires, a little extra for her granddaughter's birthday, maybe finally fixing the kitchen window that won't close right. Then she went to Gemini — Google's AI, free, gemini.google.com — and asked it to run the actual math. Not the headline. The math. Her Social Security raise, minus her expected Medicare Part B premium increase, minus the inflation she's already paying on her own grocery list. Gemini did it in 90 seconds. And what came back stopped her cold. Medicare Part B premiums have been rising about 9–10% a year. Her Part B is currently $202.90 a month. A 9.7% increase — which is what 2026 actually saw — would add about $20 to her monthly premium. That's a quarter of her raise, gone. Add her real grocery inflation (she shops at the same store every week and knows her numbers), and her actual take-home raise was closer to $30 a month, not $81. Dot wasn't angry. She was relieved. Because now she knew. She wasn't going to mentally spend an $81 raise that wasn't really $81. She wasn't going to commit to anything until October, when the official COLA gets announced. That is the whole point of doing the math now. Not to be a pessimist. To be ready. Here is the prompt Dot used. You can paste it into Gemini, Claude, or Grok and have your own personal answer in under two minutes. |
When Dot ran this, Gemini built her a clean three-column table. Low scenario: about $30 net per month after Part B. Middle: about $48 net. High: about $74 net. Same raise headline. Wildly different real-life outcomes. Then she did one more thing that I want you to do too. She asked Gemini a follow-up: "What about my drug plan and supplemental insurance? Those go up too. Should I expect another deduction in January?" And Gemini reminded her about the Part D deductible going to $700 for 2027 — which had already been finalized — and the typical 5–10% increase on Medigap plans. She added those into her forecast.
I know what you're thinking. "I'm not handing my Social Security number or my actual benefit amount to a chatbot." And you shouldn't. Read the prompt again — there is no SSN in it, no account number, no login, no personal identifier. Just the dollar figure on your monthly check, which is on the same level of secrecy as your grocery receipts. Type the number directly into the chat box. Don't upload your benefits letter. Don't paste an email from SSA. The dollar figure alone is enough math for the AI to work with. And if you're thinking "I don't know my exact benefit off the top of my head" — that's fine too. Log in to your my Social Security account at ssa.gov when you have a quiet hour, write down your gross monthly benefit and your Part B premium on a piece of paper, and use those two numbers. That's all you need.
Open Gemini at gemini.google.com. Sign in with the same Google account you use for Gmail — no separate sign-up. Paste the prompt above with your actual numbers in the brackets. Read the table. Save it somewhere. Run it again in October when the official COLA is announced. Two minutes today. Six months of clarity. |
| ★ ★ ★ |
|
| ★ ★ ★ |
|
★ Quick Wins
|
|
Skill Builder: How to Write a Prompt That Doesn't Get a Useless Answer The single biggest reason people give up on AI tools is that their first try produces a generic, useless answer. They type "tell me about Social Security COLA" and get back five paragraphs of Wikipedia-style fluff. They close the tab and decide AI is overrated. The tool isn't the problem. The prompt is. A good prompt has four parts. Who you are. What you have. What you want. What format you want it in. Skip any one of those and the AI fills in the blank with a guess — and the guess will almost always be generic. Compare these two prompts. Bad version: "What's the 2027 COLA?" You'll get a paragraph of news summary. Good version: "I'm 67, my monthly benefit is $2,140, my Part B premium is $202.90, and I want to know what a 3.8% COLA actually means for my net check in January after Part B goes up. Show me a one-line answer in dollars." Same topic. Same AI. Wildly different output. The format request is the part most people forget. AI will give you a wall of text by default. If you want a table, ask for a table. If you want bullet points, ask for bullet points. If you want it short enough to read on your phone screen, say so. The AI is happy to oblige — it just doesn't know what you want until you tell it. Here is a fill-in-the-blank template you can save and reuse on any topic, from medical bills to recipe planning to writing a difficult email.
One small habit will sharpen this faster than anything else. After the AI gives you its answer, read it once and ask yourself: did this answer my actual question, or did it answer a question I didn't ask? If it drifted, write back one sentence. "That's helpful, but I asked specifically about [the missing piece]. Can you redo it focused only on that?" The AI will course-correct immediately. It's not insulted. It's a tool. Three months from now you'll be writing prompts on autopilot. And the people around you who are still typing four-word questions and getting four-paragraph mush will be wondering how you keep getting such useful answers. |
