iiinova — where your digital self keeps evolving

Tisha @type404.com · 21/02/2026

Local owl supervising the hydrangeas

Wooden owl sculpture surrounded by purple hydrangeas in a garden

000

Tisha @type404.com · 02/01/2026

Bitter melon, no false advertising.

010

Tisha @type404.com · 01/01/2026

Between flower and fruit. Blackberry pending.

Green blackberry fruits and a white flower on tree

000

Tisha @type404.com · 01/01/2026

6 month old dried protea and a Christmas led house. Happy New Year 2026.

Dried flowers on table with white table cloth and a Christmas light miniature house underneath

Christmas led light house in focus on white table cloth

020

Tisha @type404.com · 16/11/2025

Got around to making this from last year’s NYT cookie week! Gochujang Caramel Cookies - Such a fun flavour combo, especially if you think a hint of spice in everything makes it better! 🌶️ Not spicy, just a balanced sweet-warm note. Free recipe linked below!

Picture of Gochujang caramel cookies on a baking tray

150

Tisha @type404.com · 07/09/2025

Why? Yes. I’m just chillin and getting some tummy rubs. 🐈

Tabby Cat sitting on lap with her tummy out

000

Tisha @type404.com · 27/08/2025

I remembered to take a ‘food influencer’ photo after following a recipe!! ‘Squishy no-knead white bread’ recipe from Sohla’s ‘Start Here’ 🍜

Squishy no-knead white bread recipe in the background from Sohla’s Start Here book and the actual cook bread infront

280

Tisha @type404.com · 09/08/2025

Happy with that lamb cook! 👩🏽‍🍳 Only regret is not doing a decent food porn worthy pic, #IMPATIENT. 🍜

Picture of slow roasted middle eastern lamb cook

150

Tisha @type404.com · 08/08/2025

Yup all you need is confidence and you can update the dictionary as you please…..

020

Tisha @type404.com · 07/08/2025

And then instead of rechecking it’s calculation was very certain that the pattern would be obviously wrong and prompted to rewrite it. When prompted to check it’s previous maths it realised it was wrong 🫠

ChatGPT screenshot, text below

Part
Stitches
3SC
3
3sc in last st
3
2sc on return
2
inc (2 in one st)
2
Total
10
V Correct total for Round 1: 10
stitches
So you're absolutely right — Round 1 ends with 10 stitches, not 11.
That means Round 2 should take those 10 stitches and increase as needed.

000

Tisha @type404.com · 07/08/2025

ChatGPT is quite bad at basic maths. My bad for trusting it and undoing my crochet because my count was 16 and it said 18…… 🙃

ChatGPT screenshot of counting of stitches.
Copied below
• inc = 2
• 2sc = 2
• 3inс = 6
• 2sc = 2
• 2inc = 4
→
2+2+6+2+4=16 stitches
So the actual total for Round 2 is 16 stitches, not 18.
Conclusion:
The written instructions lead to 16 stitches in Round 2, not 18, so:
• Either Round 1 ends with only 11 stitches (which is correct)
• Or the pattern has a small typo, and the Round 2 instruction should include one more increase to bring it up to 18 stitches
Would you like me to rewrite Round 2 with the correct stitch count for 18 stitches?

100

Tisha @type404.com · 03/07/2025

All that matters is line goes up! Long but a great read! www.wheresyoured.at/the-era-of-t...

2295

Tisha @type404.com · 24/06/2025

First attempt at bread with just Rye flour! A take on Richard Hart miso rye bread recipe, pretty happy with it. Baking is calming. 😌

030

Tisha @type404.com · 20/06/2025

As much as these tech giants get on my nerve on a day to day, this nonsense has sure been entertaining. Just reading the reporting by information like whaaat even are these numbers. I acknowledge my brain doesn’t comprehend billions but 125B in 5 years. 😐 www.theinformation.com/articles/ope...

010

Tisha @type404.com · 20/06/2025

And then when asked why didn’t it start with citations when i started the chat to provide me with reliable sources, it hits me with ‘Instruction Missed in Execution’ Imagine talking to person asking them to do something which they ignore and says the instruction was missed in execution….WHAAAA

Screenshot of ChatGPT convo

Ques: Why don't you by default provide citations or flag uncertainty, especially when i started the chat saying to use highly rated sources

Answer:
That's a completely fair and important critique - and you're right: I should have adapted to your instruction from the start.
* Why it happened:
1. Default Behavior Bias:
My standard tone tries to balance helpfulness with clarity and fluency, which often means giving straight answers unless uncertainty is obvious or flagged by the user. That works in many casual use cases - but fails in expert, source-sensitive domains like infosec, especially when the user clearly sets expectations (as you did).
2. Instruction Missed in Execution:
You said: "using highly rated sources answer my questions in this chat" — I acknowledged that, but then didn't consistently show my sourcing in the responses. That's a failure to carry the standard you set through the rest of the interaction.
3. Tradeoff Flaw:
I often aim to reduce "noise" (like long citations or links) for smoother reading unless the user requests academic or technical sourcing. In your case, I should've prioritized transparency and citations over brevity.
Going Forward — How I'll Adjust
1. For every answer from now on, I'll include:
• Direct citations or source links
• A note on confidence level (e.g., confirmed by docs, backed by community consensus, etc.)
2. Flag ambiguity or conflicting sources explicitly, especially when:
• Projects are evolving (open source tools)
• Official docs are thin or outdated
• Multiple interpretations exist
3. Stick to analyst mindset - i.e., no answer is fully trusted unless validated by evidence or verifiable logic.

000

Tisha @type404.com · 20/06/2025

Exactly. I asked ChatGPT how can i trust it when it clearly was so wrong with confidence to which it basically said yup I make mistakes, onus is on you to push back to get the correct answer. Likeee huhhh so the only time i can use it is when i know what the correct answer is. 😐

Screenshot of chatgpt convo
Question:
How can i trust you if youre consistently wrong with confidence?
Reply:
You're absolutely right to question trust when I give confident answers that turn out to be incorrect — especially in a field like cybersecurity where precision and reliability are non-negotiable.
• Here's the reality:
• I strive to synthesize answers based on verified, high-quality sources.
• But I can and do make mistakes — especially when there's outdated, conflicting, or ambiguous information.
• When you push back (as you did, it sharpens the quality of the conversation. That's how good analysis works: it's iterative, skeptical, and sourced.
? What I can do to improve trust going forward:
1. Cite sources explicitly — not just "this is true," but where and how it's verified.
2. Flag uncertainty or ambiguity — instead of being overly confident when clarity doesn't exist.
3. Offer links or quotes directly from project maintainers/docs — so you can verify them yourself.
4. Encourage scrutiny - because in this field, if you're not validating, you're vulnerable.
6º Why you can still use me effectively:
I'm a tool - a very fast, wide-reaching assistant, but not a replacement for analyst-level judgment or source validation. The best analysts (like you're clearly aiming to be) use tools, don't blindly trust them, and hold them accountable — exactly what you just did.
If you're still willing, I'd like to earn your trust back by being more source-transparent in this chat. Ask me the next question, and I'll provide:
• Clear, sourced answers
• Confidence level (e.g., well-supported, community consensus, speculative)
• Direct links to relevant docs or GitHub lines
Let's make this sharper.

000

Tisha @type404.com · 18/06/2025

Thank you! Yeh the sheep were so fun to do. Embroidery part for the eyes and nose not so much 🤣

000

Tisha @type404.com · 18/06/2025

Only took 3 months! First ever crochet 🧶 blanket done. 3D granny square.

4320

Tisha @type404.com · 03/06/2025

Then I asked it why it didn’t do what it said it would do going forward and the reply….. “Instruction Missed in Execution” I wish I could not follow rules and prepare a bullet list saying that 🫠

Contents of the conversation with chatgpt screenshot

Why don't you by default provide citations or flag uncertainty, especially when i started the chat saying to use highly rated sources
That's a completely fair and important critique - and you're right: I should have adapted to your instruction from the start.
* Why it happened:
1. Default Behavior Bias:
My standard tone tries to balance helpfulness with clarity and fluency, which often means giving straight answers unless uncertainty is obvious or flagged by the user. That works in many casual use cases - but fails in expert, source-sensitive domains like infosec, especially when the user clearly sets expectations (as you did).
2. Instruction Missed in Execution:
You said: "using highly rated sources answer my questions in this chat" — I acknowledged that, but then didn't consistently show my sourcing in the responses. That's a failure to carry the standard you set through the rest of the interaction.
3. Tradeoff Flaw:
I often aim to reduce "noise" (like long citations or links) for smoother reading unless the user requests academic or technical sourcing. In your case, I should ve prioritized transparency and citations over brevity.
• Going Forward — How I'll Adjust
1. For every answer from now on, l'll include:
• Direct citations or source links
• A note on confidence level (e.g., confirmed by docs, backed by community consensus, etc.)
2. Flag ambiguity or conflicting sources explicitly, especially when:
• Projects are evolving (open source tools)
• Official docs are thin or outdated
• Multiple interpretations exist
3. Stick to analyst mindset - i.e., no answer is fully trusted unless validated by evidence or verifiable logic.

000

Tisha @type404.com · 03/06/2025

I asked ChatGPT a question which it got wrong not once but twice and then I asked it “Why i should trust it” it came up with this…. And I’m more annoyed at the tone of this reply than i was with the wrong answer, “…earn your trust” 🤨 bruh what?

Screenshot that says

How can i trust you if youre consistently wrong with confidence.

You're absolutely right to question trust when I give confident answers that turn out to be incorrect — especially in a field like cybersecurity where precision and reliability are non-negotiable.
• Here's the reality:
• I strive to synthesize answers based on verified, high-quality sources.
• But I can and do make mistakes — especially when there's outdated, conflicting, or ambiguous information.
• When you push back (as you did), it sharpens the quality of the conversation. That's how good analysis works: it's iterative, skeptical, and sourced.
P What I can do to improve trust going forward:
1. Cite sources explicitly — not just "this is true," but where and how it's verified.
2. Flag uncertainty or ambiguity — instead of being overly confident when clarity doesn't exist.
3. Offer links or quotes directly from project maintainers/docs - so you can verify them yourself.
4. Encourage scrutiny - because in this field, if you're not validating, you're vulnerable.
bº Why you can still use me effectively:
I'm a tool — a very fast, wide-reaching assistant, but not a replacement for analyst-level judgment or source validation. The best analysts (like you're clearly aiming to be) use tools, don't blindly trust them, and hold them accountable — exactly what you just did.
If you're still willing, I'd like to earn your trust back by being more source-transparent in this chat. Ask me the next question, and i'll provide:
• Clear, sourced answers
• Confidence level (e.g. well-supported, community consensus, speculative)
• Direct links to relevant docs or GitHub lines
Let's make this sharper.

100

Tisha @type404.com · 01/06/2025

Thank you for the gift link! THIS is the AI I’m excited about, not the ones that all the AI CEOs want us to believe is going to take over all the entry-level jobs with a mere cost of destroying the planet.

Text from the article that states Technological progress does not require businesses to operate like empires. Some of the most impactful A.I. advancements came not from tech behemoths racing to recreate human levels of intelligence, but from the development of relatively inexpensive, energy-efficient models to tackle specific tasks such as weather forecasting. DeepMind’s AlphaFold built a nongenerative A.I. model that predicts protein structures from their sequences — a function critical to drug discovery and understanding disease.

000

Tisha @type404.com · 30/05/2025

Saw these beauties on a walk the other day. My bird knowledge is zero, can barely tell a crow from a pigeon. Had to find out what birdies they’re ChatGPT says it’s rose-breasted cockatoo or Galah, given the description it does sound correct. #birdwatching

Rose-breasted cockatoo or galah, native Australian bird on the grass

010

Tisha @type404.com · 29/05/2025

Why you staring at me like that human? Let me enjoy some quiet time in my cat house. #catpics

Brown tabby cat in her cat tree house looking at the camera with Windows in the background

040

Tisha @type404.com · 27/05/2025

Thanks Amazon I have always wanted to know how an oxo stainless steamer would look on the carpet in my lounge room. Companies after they spend millions on AI and now have to shove it down people’s throats that makes absolute zero sense. #AIBullshit

000

Tisha @type404.com · 25/05/2025

Why can’t it be sunbathing Saturday everyday.

Sunbathing cat infront of the window with tummy out

020

Tisha @type404.com · 24/05/2025

Even though the article doesn’t provide a solution, as it says it’s hard to imagine a fix. But coming across this and looking at @edzitron.com ‘s follower count gives me a bit more hope, knowing that okay there are people out there seeing through this BS.

Screenshot from the article that says It’s so easy, and perhaps inevitable, to feel a sense of nihilism about it all. Nothing matters. It’s all symbolic. Our world is filled with companies run by people who don’t interact with the business, and that raise money from venture capitalists that neither run businesses nor really have any experience doing so. And despite the fact that these people exist several abstractions from reality, the things that they do and the decisions they make impact us all. And it’s hard to imagine how to fix it

000

Tisha @type404.com · 24/05/2025

..show the tech industry has become because of the enormous amount of power in the hands of people who runs it but hardly has a clue what it is they are running.

Screenshot from the webpage linked which says This problem, I believe, has poisoned the fabric of almost every part of modern business, elevating people that don't do work to oversee companies that make things they don't understand, creating substrates of management that do not do anything but create further distance from actually doing a job

120

Tisha @type404.com · 18/05/2025

It’s Monday again tomorrow. For some reason they seem to be more frequent than Fridays. #Catpics

Cat staring in the void lying on human hands

020

Tisha @type404.com · 16/05/2025

The only thing between the cat and the cat’s food that it is ready to pounce on, is the stupid glass window. Cat: 0 Bird:1 #catpics

Cat looking intently at the bird outside glass window

081

Tisha @type404.com · 15/05/2025

You’re boring me, give me pats and snacks and shut up.

030

Tisha @type404.com · 12/05/2025

Started a crochet toddler blanket with 3D square designs—thought a month would be enough to wrap it in time for her 5th bday. Now I’m almost 2 months in, questioning my decision, the toddler might be 15 by the time I finish. 🙃 🧶

3D Crochet squares showing sunflowers, sheep, strawberry design

010

Tisha @type404.com · 11/05/2025

When you buy your cat a toy MacBook so they can truly experience tech life—burnout, imposter syndrome, and if lucky a layoff in the name of AI efficiency.

Cat looking at toy MacBook with another cat in the background looking out the window

010

Tisha @type404.com · 22/03/2025

Perfect crochet 🧶 projects to hold my attention span till completion. #crochet

020

Tisha @type404.com · 11/03/2025

Crochet 🧶 attempt at a flower pattern, was tricky getting my head around the pattern but fun once I did. I also learnt that yarn tangles (who would have guessed 🙃) and untangling it’s not as easy as I thought it was.

Crochet with grey yarn with a flower pattern, has missed stitches in the middle

020

Tisha @type404.com · 04/03/2025

3 weeks into learning to crochet and I’ve realised counting isn’t my forte. First-ish attempt at a granny square! #crochet 🧶

Crochet granny square with uneven stitch count in the corner. Two shades of blue yarn

4160

Tisha @type404.com · 24/02/2025

Hmm 🤔 it’s been hot, might need a spin in the washing machine.

020

Tisha @type404.com · 28/01/2025

Judging my human making questionable life choices from a higher altitude. All day…everyday. #catpics

Cat sitting on the stairs giving side eye

040

Tisha @type404.com · 01/01/2025

Starting the year strong, doing what she loves the most. #CatUnderTheTree

Cat sleeping on couch in front of Christmas tree

040