Brought to you by ChatGPT...

Opening Note:

As always a, welcome everyone who find their eyes on this Newsletter. This is a semi-special edition as I attended quite the event last week. The AI in Finance Summit. With that being the case, and with AI being all anyone can talk about atm. This will be an AI centric newsletter (if the Headline and photo wasn't a giveaway), and how I see it affecting the recruitment. The benefits and potential risks I see on the horizon.

The AI in Finance Summit - Takeaways:

Firstly, will just say, what a great event and there was some truly amazing and inspirational lectures and panels.

I'll keep this section relatively brief, although there is a lot to to take from the event, it was quite technical in many phases and with technical mechanics there are people far more qualified to discuss and comment than myself. However, there are some interesting takes from it.

General LLM's are not the answer, at least, not for now:

There were some themes across presentations and this was one of them, the fact that general LLMs aren't the answer to specialized problems in a sophisticated industry. You will still need to fine tune models to make the greatest impact on your business. As most businesses at an enterprise level have specific use cases that have formed over the tenure over their tenure, with their own data. Or, "MiniGPT's" as they're known.

For further information on this concept, by someone who understands it better than I do, here you go: https://www.youtube.com/watch?v=kCc8FmEb1nY

Data Privacy and Regulation may stifle enterprise innovation:

This wasn't something openly said in any of the panels or presentations, however, from having listened to all of them it was pretty blatant to me this is something that will affect their ability to stay ahead of the curve. Which, is not necessarily a bad thing.

It's so important for financial institutions to be safe, secure, and regulated, that they can't have AI/Automation running riot in their systems. Although I would imagine the automation will be sandboxed before being deployed, the big institutions will probably be a year or so behind to peak of innovation, likely having to depend on vendor products. Which is pretty much the case now.

The future is here, and it's not going to wait for anyone:

Through this summit, and my own research, there is no doubt in my mind there has been a paradigm shift and we are entering the early phase of a new way of life. The rate of adoption is going to be scary over the next couple of years, you see the signs of it now with GPT-4 allowing API access, and all the use cases springing up. Even the brightest minds in AI are having trouble keeping up with all the new developments and use cases.

There is a not so distant future where every personal device has an AI assistant on it, so fine tuned that it will legitimately be as if you have your own PA. Auto email editing, schedule management, presentation creation, formatting excels, coding automation... it's all going to be happening soon. If not already.

There is nothing this technology can not improve when fine tuned and given all the funding and research now flying into it, the only thing limiting factors are your imagination.

In the next 1-2 years everything will have some of this technology inside of it. The flood gates have opened. Don't get left behind.

How to use AI as a Candidate, looking for a job:

1) Resume Reviewal and JD Tie in: I haven't fully tested this out personally, but I've heard of multiple people doing this to great affect. Essentially, you can take your skills and experience from resume, enter it into ChatGPT, prompt it. Then add the JD of a role you want to apply for, the engine should then create a resume that ties in your experience into the specific JD you entered. In minutes, if not seconds, you have a specific Resume for the job you want.

2) Interview Preparation: This is something I've been using to help people I work with. In a similar vein to the Resume/JD merging, you can get the engine to explain to you why you're a fit for the role you're interviewing for. One of the biggest issues candidates face is their ability to sell themselves into the role and the firm. I can't tell you how many candidates I've worked with that were perfect for the role but struggled to explain why, leading to them getting passed on. This will aid you in teaching you your selling points, and what to say to the hiring firm.

How to use AI as a Client (Looking for Talent):

I won't repeat myself, but, you can inverse the above from the perspective of a hiring manager, and use it to help you select people or screen people/better understand their background before moving forward to interview. No more googling what an applicants firm does and how it fits in.

1) Explanation of your firm and the role: Something that firms can be quite poor at, is explaining to candidates what they do in words they understand. This is prevalent in tech and finance since the sector is sophisticated, the language used is very specific to it. Your job when interviewing, is to make the candidate's life as easy as possible (within reason), so when there's time to make a decision, they have all the info they need to make an informed one. You can now explain your firm in seconds, in writing, and in a way everyone will understand.

2) Write (Good) Job Descriptions: In a similar vein to the above, a lot of firms have issues sending over digestible job descriptions that properly explain the role and the division it sits, what it does, it's impact etc. It can also be quite time consuming to amend and change things, or getting senior staff to contribute to it. Now, you can spend about a minute typing a prompt explaining the firm, the role, the division, and seconds later you will have a fresh new JD.

The risks of this Technology and what to consider?

Client Data Privacy/Security:

One issue I foresee and people need to be wary of is what they input into the public general LLM's like a ChatGPT. It's an amazing tool, but every letter you put into it helps further train the model and the data is extracted back into OpenAI. With that being the case, there runs a risk of staff entering proprietary information, both of the firm or their clients, which should not be getting put into external non-approved 3rd party systems. As a firm, you would do best to start creating documentation and guidelines around this, as it will eventually cause issues as staff cut corners and enter proprietary data to get work done.

The Take Home Tech Test is done:

ChatGPT can write cleaner and better code then the average engineer in a lot of cases. Even with that aside, it is absolutely a tool people can use to write average code, and then tune the model to make the code elite. You have to keep this is mind when evaluating take home tech tests when interviewing. You will either need to start doing it live, or put in the test documentation that you expect them to use ChatGPT, and you will be evaluating their ability with the consideration they have used AI to assist them in writing it.

It still can't replace a trained eye:

I look at LLMs and AI much the same as I look at a calculator. The calculator didn't kill Mathematics, it just revolutionized our ability to do it. LLMs will not innovate anyone out of a job, except for the people who refuse to use it to improve their own performance. You still need to read over everything before putting it out, even with GPT-4 it's fairly obvious what's AI written, it has a very unnatural tone of speech for the most part. For more sophisticated and complex subject matter it is still liable to mistakes as well, which is why a trained eye will likely always be required.

A trained eye is also the only type who can prompt it to get the best out of it. You need to understand at a deep level what you want it to do, to get the correct outcome. Otherwise your prompt ability will be limited. Similar to only an person experienced in something has the ability to ask the right questions, only an experienced person will be able to create the right prompts and trade the model to optimal efficiency.

These kinds of things raise the skill floor but the ceiling still remains.

Closing Thoughts:

Firstly, as I said a the beginning, this technology is here to stay and is evolving at a truly rapid rate. In about a years time there will be an abundance of production level technology systems that will revolutionize the way nearly everything is done. From content gen, to automation, to coding. It can do it all.

What you need to start doing is familiarizing yourself with the software and how it works. It's quite simple to use once you understand the logic behind it conceptually and it it's a massive time saver of the mundane menial tasks none of us like to do. Then, when the time comes where firms start implementing this technology into their stacks, which is inevitable, you won't be playing catch-up.

This Technology is here to stay whether we like it for not.

I, for one, welcome our new AI overlords.

Thanks for reading :) I really enjoyed writing this one! - James

P.S - This wasn't actually brought to you by ChatGPT, I typed all this out myself by hand. My lord, am I boomer now?! Is this what my Grandfather felt like when he first seen an iPad?