I find myself using AI a lot. Like a lot. It’s slowly replacing Google for me. I use it to answer questions rather than trying keyword combinations first in a search engine. I usually get my answers much faster that way.
I have one rule when using AI: Trust but verify.
I check everything it says or recommends, especially if I intend to act on it in some way. Depending on any associated risks or costs I’ll check that its answer is complete and that it hasn’t missed anything. An alternative option for example.
Trust but verify.
Tag: ai
Searching for the human touch
There seems to be too much AI content. It was fun and interesting at first but the algorithms watched me consume AI content and now that’s all they seem to send my way.
On Spotify I can now identify an AI-generated music track within the first few seconds of hearing the vocals. It’s like the early days of in-car navigation systems: only a few voices were available. The few variations of male and female voices offered allows you to spot the AI singer with ease. I’m now trying to train the playlist AI to not offer AI content.
On Instagram it’s getting harder to spot the AI generated images from the real. They’ve now mastered fingers and toes and general biology. You now have to look at the shadows, facial expressions, eyes, and the overall feel of the image. Does it look too glossy? If yes, is it a filter? Also look for perfection. Perfect skin tone and musculature in every pose screams fake or manipulation. It is getting harder to tell what is real. Welcome to the Matrix.
There’s even AI videos, including short films, news articles, white papers, and podcasts all generated by AI. Ok it was cool at first but I find myself having to hunt for genuine human created content. I look for typos, grammatical errors, and human prose. I actively go out of my way to avoid AI content. Enough is enough! I want to consume something real, created with passion, interest, and human creativity.
I’m searching for the human touch.
Big faceless organisations
I miss the days before the internet sometimes. Especially when dealing with companies. They think they are being efficient dealing with customers only through the web or an app, using AI agents to deal with enquiries and having complex digital complaints procedures.
Once they have your money and you experience problems with their product or service, good luck reaching an actual human to talk to. Or getting your money back.
If you are an IT geek though you might have a chance. I’ve managed to talk to humans and get my money back on several occasions, but it took a bunch of skills that the average person just doesn’t have such as:
– Understanding how web pages work and reading the page source
– OSINT
– How to hack AI
– Side-channels
– Exploiting software bugs
It shouldn’t require a degree in IT in order to get decent customer service but that’s been my experience.
Lists 2.0
I’m a big fan of lists. To do lists, checklists, packing lists, you name it I’ve turned it into a list. I recently learned that ChatGPT is great at lists.
Very detailed lists.
Need to plan a vacation? It will generate a list of things you need to take into consideration, in addition to what to pack. Shopping for a new laptop, car, or even a house? It can quickly knock you up a list with indented bullet-points on everything you need. Need to make a plan should a disaster occur, maybe a nuclear war or zombie outbreak? Yep it will create a detailed survival plan complete with supplies list in seconds.
Lists 2.0: create your unique custom lists with AI!
Whose art is it really?
I found myself at a Christmas craft market recently, and at one stall I stopped to admire some artwork. The stallholder was selling coasters, mugs, canvas bags, and tea towels covered in her art. The colours were vibrant and the art was detailed, almost 3D.
I complimented her on her work. She said thank you then proceeded to tell me how her and her husband enjoy creating these products. Then she said “although he uses a different AI program to me”.
It turns out that she is not an artist but a Midjourney user. This got me thinking: is this therefore her art or someone else’s? Her skills lie not in creating the art itself, but in prompting the AI to create it, refining what it creates, and weeding out any errors such as 6-fingers or too many appendages.
The art was good, very good, and the prices were high, but I couldn’t bring myself to buy anything. Why? I wasn’t quite sure. Whom am I rewarding for the work? Are real artists losing out?
What are your thoughts on works created by AI?
ChatGPT hacking buddy
I’ve given many talks on cyber security and ethical hacking over the past few years and one of the things I tend to say a lot is: “There is no such thing as cheating in hacking”.
What I mean by this is that as long as you are learning you are not cheating. It’s only when you take shortcuts and learn nothing in the process that you are cheating yourself.
So when attempting a CTF or Hack the Box or Try Hack Me machine and you get stuck and you have exhausted every technique and trick that you know and nothing is working, sure, go search for a writeup or forum posts on how to progress. Read just enough to get yourself unstuck and then keep going. Learn the technique, tool, or whatever you needed to know to progress. Add it to your knowledge-base.
This is learning not cheating. Finding the answer but not learning how and why it worked is just cheating yourself.
Recently I found myself stuck on a CTF that I was taking part in for fun. It was brand new so there were no writeups or forum articles to peek at. And I was stuck. In theory I could just move on to the next challenge and come back to this one later, time allowing, but I was having fun and I wanted to figure out why my solution wasn’t working. I wanted to learn, now.
I decided to see if AI could help. I’ve been playing around with the free version of ChatGPT recently and wondered if I could make use of it in this situation. I gave it a copy of the code from a program I had disassembled as part of the CTF and asked it to tell me what the code was doing. It did, in great detail. I then asked it how I could extract certain data that the program was storing in memory. It gave me detailed instructions using a tool that I was unfamiliar with. I asked if if I could do the same with another tool I was familiar with. It kindly said no and offered to teach me how to use the tool it recommended. I agreed and learned how to use the tool and managed to make progress.
I then continued hacking at the CTF asking ChatGPT for assistance when required. Although technically cheating, I was constantly learning throughout, and allthough I managed to get some virtual points on a virtual scoreboard, they were worthless in the real world, but the knowledge I gained from hacking with ChatGPT was priceless.
So now when I get really stuck and I’ve exhausted everything I know, I turn to ChatGPT as my AI hacking buddy. Only after I’ve finished the challenge, or both ChatGPT and I have failed to come up with a solution do I go looking for a writeup.