In AI Standing Still is Falling Behind - Learnings about best using AI from London Tech Week

Learnings about best using Generative AI for #bionic from #londontechweek

?

Upon walking down to Tobacco Dock on Wednesday I was struck by the number of people trying to get in, it wasn’t quite Glastonbury but there was a queue round the block. Then, inside I was taken aback by the number of people there from all over the world and the general buzz in the air.?The beautiful London Summer’s Day helped but it’s hard to remember a feeling like this at a conference.

?

I was there to learn more about Generative AI in person to figure out how we can use it at #bionic .?

For example could we use #chatgpt to better empower human like conversation????

?

Learnings

?

Everyone thinks that #generativeai is a game changer.?The opening remarks mentioned that Analytical AI has been around for 15 years but this focuses on analysing existing data for predictions and automation.?Generative AI, however is a different animal for 3 reasons:

1.?????This new type of AI learns from data to create new data based on what it learns.?

2.?????This new data also looks very much like something a human would produce.?

3.?????It is so simple to use that you don’t need to be techie to use it, anyone can create data.?

?

Why it’s happening now:?basically it's due to increased computing power and new faster models (transformers). These new generative models need vast amounts of diverse and messy data.?It's only now that we have the necessary processing capability.

?

So what can it do:?Much of the discussion was around using it as an assistant – Microsoft’s CoPilot platforms seem aptly named – basically a person prompting it to augment existing processes. In this way it could be used anywhere you have repeating tasks.?But, here I picked up three commonly mentioned use cases:

1.?????Content generation (e.g. generating copy);

2.?????Content collection and summarisation (in real time) and;

3.?????Semantic services (a chat bot communicating in the style of a person).??

?

What is responsible AI?:?It may have been coincidence that the EU passed its inaugural AI Act on the first day of this global AI summit but discussions around governance, security and ethics were prevalent in all sessions.?From concerns about bias in the models and importance of diversity in the inputs, training on the wrong data, understanding/trusting the black box models, problems defining IP to general ethics:?is AI going to power social inequality? Are we sure we know what we are opening up here??There is an incredible amount to be thought through.

?

Recommendations/Opportunities

?

What’s the best approach for an organisation that wants to start using it:?In considering using generative AI I learnt about; the importance to build in data governance and defined structure around models, to not build your own stuff, start small and get buy in through proving value cases. Oh, and start now or it will be harder to catch up!?

?

What are we doing here at #bionic : we are working with an expert partner (#Quantspark ) and liaising with other companies in our investor portfolio to explore how we can use AI to help us accelerate our digital transformation and the business in general. We have organised a series of workshops to understand any user cases and the necessary inputs. Building on the learnings we are looking to start small with an existing process and put governance frameworks and policy in place as we do so.

For more information please don't hesitate to get in touch.

?

Other take out: ?Coming from a #data and #dataanalysis background I found it very interesting to attend a conference where much of the content is about adjunct material. It’s really added to my thinking.

?Perhaps ironically #chatgpt was not consulted in the writing of this article..