Generative AI. Speedy. Efficient. A potential game changer. The newest content creator on the block and possibly your new(ish) best friend if you’re trying to whittle down your to-do list. Without a doubt, generative AI brings a wealth of benefits with it which enterprises can take advantage of. Its efficiency paves the way for increased business productivity, speedier analysis of data, better and faster personalisation plus faster R&D processes. It can also be used to formulate new business models that could help you revolutionise your way of working.
While these are all huge advantages, situations like Samsung’s recent unsuccessful dalliance with generative AI (see here for more), have reminded enterprises of the need to proceed with caution. Yes, it is very exciting that your newest AI buddy may be able to help your workforce whizz through pending tasks but how can you as an enterprise ensure that you mitigate for AI faux pas? In plainer terms, how can enterprises manage generative AI in a responsible way?
In order to consider this, let’s first look at the lay of the generative AI land. We’ve already covered some of the benefits – efficiency, speed and all that jazz – but what else do you need to think about if you are to take full advantage of these?
To start with, you have to be wary of what you input to public generative AI tools. Not everyone takes care with the prompts they use and could unintentionally (or, misguidedly) share information that the models (especially those of the public consumer-facing, non-Google variety!) shouldn’t have access to. This could lead to some unneeded data leakage situations. In cases where sensitive information is involved this could be detrimental. It could be that your sensitive customer data gets out meaning you could find yourself on the wrong side of GDPR – a situation no enterprise wants to find itself in. Or it could be the secrets of your latest innovation-to-be meaning that you potentially lose an advantage over your competitors – far from ideal. Regardless of the type of information, the bottom line is that you have to take great care with what you input to public generative AI tools. Once your information is in the model, that is it – your data has joined the model potentially becoming the world’s best kept secret rather than just your own.
You also have to be wary of the output from generative AI tools. Owing to various reasons, enterprises must keep in mind that output cannot be taken at face value. While the information that comes out may sound valid, that may not always be the case.
One of these reasons is to do with intellectual property. AI-based language models like ChatGPT draw upon publicly-available information to generate their own responses. While efforts have been made to limit their use of copyrighted material and to have such models draw from legal and ethical sources only, it’s hard to completely eliminate usage of this material as it is in the public domain.
Other reasons include the presence of toxicity and bias on the internet. Generative AI models learn (or “learn”) from all that the internet touches. Unfortunately, this means that alongside all the positive aspects of the internet, there is still toxic rhetoric and all sorts of biases available to inform a chatbot’s output. You’ve also got to be wary of issues such as hallucination (where AI generates information that seems plausible but is actually false due to inconsistencies or outright fabrication) or outright misinformation (think deep fakes where AI has been used to digitally alter images and videos of people so that they convincingly appear to be someone else). On top of all of these, you’ve also got to contend with the fact that inputting the same prompt on different occasions may result in very different outputs in each instance.
While what we’ve pointed out may have caused a frown line or two upon your face, they shouldn’t be things to scare you and your business away from generative AI – they are simply things to think about when working out how to manage this exciting technology. They also point to the fact that, exciting as it is, generative AI can’t do it all (at the moment!) so bona fide human professionals are still very much needed to ensure that enterprises stay on the right side of it. Plus, with innovations such as miniaturisation on the horizon, we are going to need bona fide human input to help us take the best advantage of generative AI far into the future.
One area in which bona fide human input will be very handy in getting the most out of generative AI is in the emerging field of prompt engineering. This recent concept is all about understanding what makes generative AI tick and how to get the most out of it. Or, more precisely, which carefully crafted prompts will allow you to get the information you’re looking for out of generative AI. We’d say that prompt engineering is going to be an essential resource for enterprises and is something that will help to create better AI that will serve to enhance the benefits of speed and efficiency that we are seeing today.
If you’re on the lookout for generative AI that works for you and your business, we won’t beat around the bush – look no further than Google Cloud. We say look no further because their generative AI offerings truly understand and address the needs of enterprise. Their generative AI models champion accuracy of data and minimal hallucination plus they whittle out toxic and harmful content. Privacy and the ability to control your data are also key considerations allowing you to safeguard appropriately. For example, platforms such as Vertex AI allow each tenant complete privacy, ensuring that their input does not get mixed in with other tenants and also the wider large language model underpinning the platform. Offerings like this work to serve enterprise rather than weighing it down with unneeded controversies that have infamously plagued consumer versions. Business is top of mind which means you can rest easy knowing that you’re in safe and trusted hands with Google Cloud.
Ancoris, as your friendly neighbourhood (and multiversal!) Google Cloud Partner, is here to help you take full advantage of Google Cloud’s generative AI offerings. We work day in and day out with AI so we are well-positioned to help you navigate the twists and turns of generative AI and the wider AI landscape plus we can keep you clued up on best responsible AI practice. To support you further, we also have a newly-established service that is specifically designed to help decision makers and teams cut through the ‘hype’ that comes along with technologies such as generative AI. This new service aims to help you get to grips with what these new and leading edge technologies mean in real terms for your skills base, governance and operating models so that you can stay ahead of the game.
So, all in all, generative AI is an innovation that is worth the hype for enterprise but there are certainly some things that you need to think about if you want to integrate it into your operations effectively. There’s nothing to panic about though, Google Cloud has heard your business needs and has you covered. And, of course, we are always here to help. If you would like to find out more about how we can help you manage generative AI and also other hype techs on the horizon, please get in touch with us today.
P.S. If you’re wondering whether generative AI has authored this piece, the answer is no. This article is brought to you by some bona fide Ancorians – we promise 😁