Clayton Redford’s Post

While LLMs are massively beneficial and are a huge benefit to someone who already has specific knowledge or is learning something new, using them to spit out answers without understanding creates a huge risk of not being able to validate the output and spot/correct any errors or hallucinations. Learning is about making and overcoming mistakes. That process cannot be shortcut.

Yup, that's me at 3:10am right now. I just resubscribed to Hack The Box and I'm absolutely addicted to this Penetration Tester learning path... Currently learning about performing recon on nameservers and mail servers. Most other people would be sound asleep at this point but I'm in my absolute element! Having done a few basic courses on Network Security, the approach from HTB is just SO my cup of coffee. It's an intense mixture of bite-sized information sections with lots of small task-based practical activities to go along with it. I'm combining this with using Anthropic's amazing Claude AI as my personal NetSec tutor to bounce questions off, which puts the amount of data I retain through the roof! (also, ChatGPT is an absolute woos that didn't feel comfortable discussing pentesting methods with me...) Might get some sleep, might not... (although if I do, I'll be dreaming of nmap scans 🤣)

Two ChatGPT prompts to consider, taken from Enrollify's guide to prompt engineering. If X, then Y "Give me 3 options. If I choose a number, expand on that option in more detail with bulletpoints." For each X, do Y. "Create 3 student personas for a Graduate Program in Artificial Intelligence. For each persona, give a name, location, income, psychographic profile, and three quotes in the person's own words about their biggest challenges. Then explain which persona we should focus on, and explain your reasoning."

I used to be a developer (I still write some code for fun, from time to time), so I love to keep up with how software is eating the world -and AI is eating software. #AI is a great tool to accelerate, augment, and automate almost everything. But it doesn't discriminate, it can also accelerate, augment, and automate your screw-ups, too. A researcher Bar Lanyado published on the Vulcan Cyber site his findings showing that ChatGPT-generated code tends to hallucinate calls to packages (code libraries) that don't really exist. This opens the possibility of someone discovering this vector of attack and then creating their own package with their own malicious instructions to be downloaded and installed without suspicion. Augment, don't try to automate everything. Keep the human in the loop, people. (link to the article in the first comment)

Hello to my network! I just completed A Prompt Engineering course for ChatGPT through Vanderbilt University in partnership with Coursera. This Course covered: 🌟 multiple patterns and how to use to be able to produce predictable outputs ✨ taught me the basics of how LLMs operate behind the scenes 💫 as well as touched slightly on NLPs and how they can affect the outputs Huge thanks to Vanderbilt University and Coursera for providing this course Also a big thanks to Jules White for providing his knowledge this is the first step of many for me to be able to learn as much as I can about AI

Enhancing Domain Security with ChatGPT: Training Models and Validation through Reinforced Learning

👉Ask ChatGPT: Give a summary of what is 'SECURE BY DESIGN' for tech products - Part 2 of 2 REF: Secure Development Practices: Developers follow secure coding practices and guidelines to write code that is resistant to common security vulnerabilities, such as SQL injection, cross-site scripting (XSS), and buffer overflows. This includes input validation, output encoding, parameterized queries, and secure error handling. Continuous Security Testing: Secure by design incorporates continuous security testing throughout the development lifecycle, including static code analysis, dynamic application security testing (DAST), and penetration testing. This helps identify and remediate security issues early in the development process, reducing the likelihood of vulnerabilities being exploited in production. Security Awareness and Training: Developers, engineers, and stakeholders receive security awareness training to understand their roles and responsibilities in maintaining the security of the product. This includes training on secure coding practices, threat intelligence, incident response procedures, and compliance requirements. Overall, secure by design is a holistic approach to cybersecurity that emphasizes building products and systems with security as a foundational principle. By integrating security considerations into the design and development process, organizations can create more resilient and trustworthy technology products that better protect against cyber threats and vulnerabilities.

By the end of this course, I am be able to: Write effective prompts to interact with ChatGPT. and Apply prompt engineering techniques to solve real-world problems. Enhance productivity and creativity using ChatGPT.

Completed this amazing course on Prompt Engineering for ChatGPT from Coursera. A well-planned course that taught me how to write prompts in an effective way to get the desired solution.

Great training in ChatGPT prompt engineering, delving into crafting prompts to address diverse patterns and tackle complex tasks head-on with tailored solutions.