(Please excuse the crossposting).

I just wanted to give everyone an update on what’s new in the world of AI. I think I can safely say that last week was the most innovative week in history. 

On Monday Runway Research introduced Gen-2. This tool allows people to realistically and consistently synthesize new videos. They can do this either by applying the composition and style of an image or text prompt to the structure of a source video (Video to Video). Or, using nothing but words (Text to Video). So basically you can take one video and turn it into a different sort of video (think cinematic to animated) or you can describe your video in the same way that you can describe an image that you want to create. 

Tuesday saw the introduction of Google’s Bard product, which they hope will rival ChatGPT. I have played with it a lot and I have to say that OpenAI and Microsoft have absolutely nothing to worry about at this stage. The description that I used on Twitter was ‘if ChatGPT is Blackadder, then Bard is Baldrick’. It is really very poor, hallucinates far more than ChatGPT, can’t even understand how to construct a basic Google search using Google’s own syntax and it is just poor. Also on Tuesday Bing produced its own image creator. It’s not perfect, but it’s completely free to use. I think we can expect to see a lot more of this happening in the future as the big companies start introducing AI elements into their products – we already know that Microsoft, Meta and Google are going down this route. 

On Wednesday Opera introduced contextual AI prompts and sidebar integration with GPT based services. Users can now expect to see smart AI prompts accessible directly from the address bar, or by highlighting a text element on a webpage. Users can use the AI services to shorten or explain articles, generate tweets or get other related content. This is another example of the way that AI is being integrated directly into the tools that you’re already using. 

Thursday saw what for me is the biggest innovation of the week – ChatGPT has introduced the use of plugins. These are tools to help ChatGPT access up to date information (a current weakness), run computations and use third party services. This already includes links to resources such as Expedia, Shopping tools, Open Table, Wolfram Alpha and so on. It’s at this point that things get very interesting. Say for example I want to take a holiday to Canada. I can ask ChatGPT to give me some suggestions on where I can go and what I can see, based on the discussion we have on my interests. I can then get it to interface directly with Expedia to book my hotel, get restaurant recommendations and a direct link to make reservations. What’s missing from all of this? The need to go to any of the apps that Apple or Google supply. Everything is blended together in a single tool, seamlessly. ChatGPT at this point is becoming the martini of the computing world – anytime, anyplace, anywhere. There is also a retrieval plugin, which allows you to upload your own data sets and use ChatGPT to retrieve information from this data. You’ll be able to upload all of your emails and notes and have ChatGPT summarise everything for you, or answer questions based on your own content. These plugins are currently in beta testing, but I can’t imagine it’s going to be very long before we will be able to make use of them directly from ChatGPT.
On Friday it was confirmed that ChatGPT4 has one trillion parameters, while its predecessor, GPT3 has 175 billion parameters. The number of parameters is an indicator of its ability to understand and recognise complex patterns and relationships in data and to develop new emergent capabilities. In short – it just got a whole lot cleverer. 

So – that’s what happened last week. It’s virtually impossible to keep up with what’s going on, but hopefully that’s given you a bit of an overview.
Meanwhile my course on AI tools for Information Professionals is taking place this Wednesday, between 12-1. I’m going to try and make some sense of what’s going on, and give you practical examples of what ChatGPT can do for you in areas other than search. I’ll also look at how you can use AI to create presentations, how you can use AI to write content, summarise material that’s already been produced, video creation using avatars that you have created and anything else that I can pack into an hour! If you’re interested there are still spaces available at https://www.eventbrite.com/e/ai-tools-for-information-professionals-a-power-hour-with-phil-bradley-tickets-546248242197 I can almost promise you that it will be the best £35 you have ever spent, unless you’ve been on one of my other courses, in which case it will be just as good!

########################################################################

To unsubscribe from the LIS-LIRG list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=LIS-LIRG&A=1

This message was issued to members of www.jiscmail.ac.uk/LIS-LIRG, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/