top of page

GenerativeAI: what are businesses actually doing?

Suitable for anyone interested in AI trends


It was great to see so many of you at our online follow up on GenerativeAI last week and a special thank you to our speakers Kate Janssen, Anna Korhonen and Haibo E and many others for sharing your expert insights. Top 10 takeaways are below:



It was a fascinating discussion on what businesses are actually doing 9 months on from the launch of ChatGPT, as well as sharing the trends our tech leaders are seeing and hearing some great tips on the future of GenerativeAI from within our community.


Here are our Top Ten Takeaways
  1. 🧐 GenAI is not the only AI solution 🧐The emergence of GenAI has reopened the question for many companies of "how are we using AI?" For some companies, what is most valuable may not be generative models, but more established machine learning approaches. For others, this new wave of models that generate new content, dangles the promise of major business wins

  2. 🏆 GenAI is giving companies gains in some areas 🏆Seeing engineering teams achieving very good productivity gains, experiments in customer success / service are making teams more efficient, day-to-day ad hoc productivity gains for execs / teams using AI for summaries or getting started (the "blank page" problem) 

  3. 🤯 BUT generally it's harder than many seemed to anticipate 🤯"It's easy to build a prototype or demo but really hard to build a scalable, safe, robust Machine Learning system". "The mass corporate use of the original products, especially for customer facing products, still has issues of reliability of output, data privacy and then spiralling token costs once you're hooked on it". Academics are still working out "whether the models can even be applied to real world issues". Not really understanding how the models work can lead to legal risks. Lots of functions have to get involved which can be hard to coordinate.  

  4. 🔨 Often feels like there's an AI hammer in search of a nail 🔨There's a lot of hype and pressure (particularly from investors) for companies to implement AI, even without a solid understanding of the customer problem. Most CEOs are still working out when using GenerativeAI is a competitive advantage vs a huge distraction. And who has the technical ability in their org to answer the key questions on what AI approach is right for the business? 

  5. 🧩 Data foundations are fundamental 🧩Until you have these, AI is NOT the solution! Generally skipping steps in data collection, organisation, cleaning, structuring is going to lead to poor outcomes. 

  6. 💸 Watch out for legal issues and high costs 💸From the risks of privacy, copyright and other legal issues to high costs of consultancy and spiralling costs of processing and tokens at scale, GenAI can lead to risks and costs businesses will need to manage. Watch out for the cross functional considerations which can be easily overlooked. 

  7. Regulation is coming ✅Regulation is trying to catch up with the growth in the area of AI and may soon change what is legal / possible in the UK. Responsible AI regulation is expected in the UK in October but there is an important summit on AI Safety being held in the UK in early November where companies hope to get more clarity. 

  8. 🌎 Seek to augment rather than replace your workforce 🌎We've seen junior developers progress their skills to senior level at record pace through pair programming with Gen AI but what about other roles? Customer service orgs seem to be at highest risk of downsizing and the question remains: how do companies balance driving adoption with the implications for downsizing and reskilling? Given the high price that could be paid in trust if customer-facing things go wrong (or high price of tokens if used at scale), advice is to start small (eg using AI for triaging) and use it to augment your workforce rather than replacing. 

  9. 🔮 AI-first organisations: By the end of the decade...🔮There will be companies in all sectors which are strongly AI-enabled / AI-first, meaning they will be using AI/ML both in many of the organisation's internal processes and ways of working and/or using it in their customer facing product, where it makes sense. So existing companies need to keep making progress to not lose competitiveness where AI works well. A good understanding of the capabilities, benefits and limitations should help guide its use cases in organisations. Importantly, AI cannot solve every problem at the moment and we shouldn't force or expect it to.  

  10. 🏃 Crawl - walk - run 🏃For established companies, starting with some easy internal wins whilst figuring out the bigger customer facing problems which are extremely complex seems to make sense. For AI-first companies, funding isn't a given - investors are definitely looking for proof points and competing against the might of the internet titans who can invest in this area is hard.

 

🚀 Thank you again to our experts who steered our discussion.

We'll continue to follow the trends in future events 🚀

 

4 views0 comments

Comments


bottom of page