#16 - Braiding the Future

Sponsored by EVA.ai

EVA.ai, a pioneering force in the HR technology space, is dedicated to transforming Talent Acquisition, Talent Management, and the broader HR landscape through its Conversational AI solutions. Embedded with advanced AI capabilities, EVA is meticulously designed to streamline the entire talent lifecycle, from sourcing and hiring to employee engagement and development.

EVA is built upon a vast conversational knowledge repository (100+ topics) encompassing all facets of HR including recruiting, onboarding, skill profiling, talent development, mobility, dynamic assessments, pulse surveys, and much more. This enables EVA to offer personalized, AI-driven recommendations and insights that empower HR professionals to make informed decisions, enhance candidate and employee experiences, and stay ahead of talent ops.

EVA has recently announced their 7.1 release, a truly matured phase of innovation with Large Language Model (LLM) and a beautifully conceptualised skill taxonomy management that has ripple effects on hiring, talent mobility and workforce planning automation. You are invited to explore the transformative potential of Conversational AI with EVA.ai team.

Head on to the link and ask for a demo: https://eva.ai/ai-conversational-assistant-hcm-hr-recruitment-experience-hiring-people-engagemen-nlp-nlu-chatbots/

Hello H.A.I.R aficionados

Yes, it's another edition of H.A.I.R and another punny title. I've got plenty more where that came from ;-)

Last week saw me traversing the United Kingdom as I headed out to various events - IHR - In-house Recruitment #BHAMLIVE24 and Resourcing Leaders UK Recruitment Leaders Summit were the highlights of my week.

At the latter event, I was invited to present to a room of TA Leaders on (obviously) AI in Recruitment. The brief - "none of that ChatGPT prompting crap, we want something tactical that we can take away". So, I gave the audience an overview of how to prepare for AI Transformation.

One of the big questions I get asked over and over again is how do TA leaders ensure that their technology partners are not just selling Snake Oil. So, I presented 12 essential questions to ask your AI Vendors.

As regulations come into play, the cost and penalties for irresponsible use of AI will soar, so the key to navigating this complex field is understanding what to look for and the right questions.

The following 12 questions can help ensure your AI vendors are managing things responsibly:

Next week, I'll give examples of good and excellent answers...

EU AI Act final vote moved to 13 March

The European Parliament's plenary vote on the proposed AI Act was moved up by one month to 13 March, Bloomberg Law reports. Following the final vote, the AI Act would enter into force 20 days after publication in the official journal.

Using AI in Employment/Recruitment is deemed as High Risk within the Act, so we need to get our heads around it if we are going to be carrying out AI-powered People-related activities in the EU.

Reviving the Metaverse: How Generative AI Ignites a New Era of Immersive Technologies for HR

Despite scepticism around the metaverse, we're entering a new era where generative AI is driving the resurgence of immersive technologies. This innovative leap forward dispels the notion that the metaverse is "dead." Instead, it highlights a renaissance, with AI at the helm, crafting more dynamic, interactive, and personalized virtual environments. For HR and recruitment leaders, this means exploring uncharted territories of employee engagement, training, and global talent acquisition, making the future of work not only a digital extension of reality but a playground for limitless possibilities and innovation.

Today, things feel different. Sora generates complex video scenarios and Groq's deterministic chips reduce machine-learning algorithms for faster processing. Together, the convergence of these advancements pushes us closer to achieving real-time video simulation and true virtual reality.

The World's First AI-Powered Job Mapping Tool to Track the Creation of AI Jobs Announced by UMD-LinkUp

Exciting news for those following the growth of AI in the workforce! Researchers at the University of Maryland have developed a groundbreaking tool called UMD-LinkUp AI Maps. This innovative solution promises to revolutionise the way we track AI jobs.

Traditionally, identifying AI-related positions has proven challenging. UMD-LinkUp AI Maps cuts through the noise by leveraging artificial intelligence itself! The tool utilises a large language model to pinpoint genuine AI jobs from general job postings.

This development holds immense value for HR and recruitment professionals. With UMD-LinkUp AI Maps, you can gain a clearer picture of the AI job landscape, understand where these roles are concentrated, and anticipate future trends.

As workers embrace AI, employers are slow to issue guidance

A survey from software vendor Slack found 43% of employees in companies that use AI have been given limited guidelines regarding the technology, ComputerWorld reports. The research also found that 81% of executives "feel 'some urgency' to roll out tools within their organizations."

Claude 3 is Amazing!

Claude 3 was released last week, and I've been playing with it pretty solidly ever since.

If you've ever chatted with me, you'll know that I’ve always been a big fan of Claude. For writing tasks, Claude tends to be the better option than GPT, because of its more natural, human-like tone (which is a great start before feeding it your personal communication style).

Now they've released a new version, and according to benchmarks, this one is even better than GPT-4, which no other model has accomplished so far. But benchmarks are one thing, the true test is what the users are saying.

Here's 3 things that I've learned so far:

Claude wants you to use XML tags when prompting it

On Claude's official website, they recommend using XML tags when prompting Claude. Here's what they say:

I've tried this and have noticed that it indeed helps Claude to better understand the different parts of your prompt, especially when your prompts get larger.

For example, here’s my one-shot prompt to create any new recruitment marketing asset, adapted for Claude with XML tags:

<INSTRUCTIONS>
- Your task is to rewrite a [asset] for me.
- I will first provide you with CONTEXT about my [role, company], about me and about any other relevant information.
- I will then provide you with a copy of a ORIGINAL [asset] by another creator.
- Your task is to take the ORIGINAL [asset] from the other creator and rewrite it for my purposes using the CONTEXT about my [role, company] and other information I have provided.
- You must use the same structure, style and tone of the original [asset]. The resulting [asset] should resemble the original [asset] very closely, but it should be relevant only to the context provided by me.
- Also, Instead of generating a link, when it's actually an image, just write [image]
</INSTRUCTIONS>

<CONTEXT>
[Enter your CONTEXT. This can be as detailed as you want and have multiple sections.]
</CONTEXT>

<ORIGINAL ASSET>
[Paste the ASSET, i.e. landing page, email, video script etc]
</ORIGINAL ASSET>        

The context and asset parts can get quite large and without XML tags this could get lost in the rest of the prompt, leading to inaccurate results.

I’ve seen this happen with ChatGPT before, and a large prompt can cause it to get confused.

Using XML tags helps Claude keep the instructions, context and original asset sections separate in this case and better understand what to do.

Handy use case of the vision capabilities

Claude's vision capabilities are pretty incredible.

So far, I've mainly been using it for extracting text from images, but I’m sure there are many other good use cases too.

Here you can see it in action. I’m pasting an image of a recipe and simply asking Claude to extract the text from that recipe.

As you can see, it's extremely fast. So whenever I need to get text from an image to paste it somewhere or just save it into my notes, this is super useful.

I’ve tried to do the same thing with ChatGPT several times before and it usually either fails and times out, or takes way too long.

Large context window

The large context window is another of Claude's superpowers.

Right now, at launch, the context window of Claude is at 200k tokens and will be extended to 1M tokens soon, according to their announcement. (Google's Gemini Pro has recently also opened beta access to their model that can process up to 1M tokens). In comparison, ChatGPT turbo currently sits at just 128k tokens, while the GPT-4 model available in the chat interface is only capable of 32k.

And trust me, a large context window makes a huge difference.

Not only can you now upload huge amounts of data and reliably process the information in it, but your chat conversations can also go on for much longer and no information will be lost.

Now the only reason Claude can't fully replace me and write this newsletter all by itself is that it still struggles to generate content with real substance.

It’s not capable, yet, of coming up with original ideas and insights when I tell it to “write a post about Claude”, as I did here. E.g. It wouldn’t have come up with the tip to use XML tags, or that the vision capabilities are really useful for extracting text, as I did.

But here's the thing.

Claude is already a perfect human-level EDITOR.

It does an amazing job of turning my thoughts into polished, engaging language that sounds just like me.

I just need to feed it my ideas, do some smart prompting and let it turn my ideas into words.

Regulation Updates

Keeping this section going as there's been great feedback on the usefulness of keeping up to date with regulations...

UK publishes responses on public entities' approaches to AI governance

The U.K. government's Committee on Standards in Public Life published the responses from government departments and public bodies on their governance processes for AI. The initial request for responses was issued in July 2023.

OECD publishes report on data scraping ethics

Duke University Science and Society Fellow Lee Tiedrich wrote for the Organisation for Economic Co-operation and Development about the ethics of data scraping for AI systems. Tiedrich explored the controversies around the practice and what solutions have been studied by international regulators.

Garante alleges OpenAI's ChatGPT violated GDPR

Italy's data protection authority, the Garante, found OpenAI's AI chatbot ChatGPT breached several provisions of the EU General Data Protection Regulation following a fact-finding inquiry. OpenAI has 30 days to submit counterclaims.

I'm Martyn Redstone - founder of PPLBOTS, AI training and consulting for Recruitment and HR Leaders, and founder of Bot Jobs, the leading job board for Conversational AI roles globally.

I'm always here to chat about all things HR, AI, and the future of recruitment. If you have questions, insights to share, or just want to connect, drop me a line. Let's continue this conversation!

Martyn

Sponsor H.A.I.R - Putting the AI in HR

Are you ready to shine in the world of AI-driven HR and recruitment? Have a product that you want to show off? Join us as a proud sponsor of H.A.I.R, the go-to newsletter for all things conversational and generative AI in HR.

Why Sponsor H.A.I.R?

🚀 Reach a targeted audience: Showcase your solutions to HR professionals, recruiters, and industry leaders passionate about AI.

🤝 Build lasting partnerships: Collaborate with thought leaders in AI and HR, forging valuable connections.

📈 Boost your brand: Elevate your AI offerings and position your company as an industry innovator.

📣 Amplify your message: Gain exclusive visibility in a newsletter dedicated to HR AI insights.

Contact me today to explore sponsorship opportunities.