Auto-GPT is ChatGPT’s younger, more performant sibling. It was released by OpenAI on March 30th, 2023, by the same company that founded ChatGPT. But Auto-GPT is vastly different. An AI agent, or rather, a multi-agent system, Auto-GPT works by autonomously completing entire goals. Essentially, in being made up of multiple systems, each with a task of its own, Auto-GPT has a certain degree of autonomy when it gets given a task. It doesn’t need any further instructions besides the initial guidance it receives from the user. But how?
ChatGPT’s chatbot system requires human input after every stage of its ‘thinking’ – it is, after all, an interactive model. Auto-GPT, on the other hand, has an automatic research loop programmed within it. This means that, given a certain goal – which can be quite complex at times – the system breaks it into sub-tasks, which it then hands over to GPT-4 and GPT-3.5. These are two simpler subsystems which can easily communicate with Auto-GPT. Using this automatic loop, GPT-4 and GPT-3.5 can then produce data for Auto-GPT using the internet. Auto-GPT can then focus its abilities on memorising this input and assigning the next objective for the completion of the overall goal. Sounds complicated, right? Here’s an example:
1. We gave Auto-GPT this overall goal: ‘Get the best solicitor training contract.’
2. Auto-GPT then divided it into three initial tasks: ‘Identify top solicitor firms offering training contracts’, ‘Analyse and compare the training contract programmes’ and ‘Optimise application materials based on their specific requirements’.
3. After writing an initial result summary for these three tasks, Auto-GPT assigned itself a ‘next step’ objective: ‘Apply to Clifford Chance for their solicitor training contract program and optimise application materials based on their specific requirements’.
In conclusion, Auto-GPT was able to find out what a solicitor training contract is, what the top firms offering them are and what their specific requirements entail. It then autonomously picked Clifford Chance out, researched their programme requirements and offered detailed instructions on how to apply and, statistically achieve, a solicitor training contract at their firm.
But Auto-GPT is only a very new model, and it has its issues. One of them is confabulation. The vast majority of the information stored online was created through human input, which of course is at least somewhat influenced by subjective experiences and values. These human traits can confuse the AI system, resulting in it misinterpreting or miss-assigning data, which it then stores and uses for its next task. This can then result in AI ‘hallucinations’, which is when an AI response is very confidently wrong.
Find out more about the essential digital skills you’ll need as a future lawyer.
If you are not a fan of the recent issues raised by ChatGPT, Claude might be your answer. Claude is similar to ChatGPT in many ways – it is also a virtual assistant, which helps with research, summarisation, coding and more. But it is ‘much less likely to produce harmful outputs.’ This means a couple of things; firstly, due to there being precursors using similar technology, Claude generally makes fewer mistakes. Secondly, Claude is much more conversational than a lot of its competition, since it can ‘take direction on personality, tone and behaviour.’
Claude was actually developed by Daniela and Dario Amodei, two former OpenAI researchers. They founded their own company, Anthropic, in 2021, after leaving OpenAI due to ‘directional differences, and their company ethos is based on responsible AI usage.
Gamma is an AI-powered application you can use to create working presentations, decks, documents or even web pages in under a minute. Gamma’s presentation software is varied, allowing for the development of your unique style for any slides or decks you are wanting to build. Gamma also allows for collaborative projects, and its built-in analytics allow users to easily measure post-engagement.
The app allows you to ‘focus on your copy’, rather than on the design and overall look of your work, which a lot of people find tedious.
With Gamma, users can focus on the quality of their information and ideas rather than on any aesthetic aspects. This kind of technology can seamlessly speed up processes, making it a very attractive feature for businesses.
Rose is a cloud data platform designed to help its users visualise data sets. It can integrate external data, which it can resource online, with internal data input by its users. Based on this data, Rose can then create infographics and charts to help you visualise this information.
Rose is especially useful for visualising and integrating big data sets in short periods of time, making it particularly applicable to the field of finance – which would be its primary market. Rose is able to interpret questions in order to correctly frame and analyse its research, using mathematical logic. Therefore, it can answer difficult technical questions in a short matter of time by producing both written and visual explanations.
AI is becoming increasingly relevant in a large number of industries, from finance to marketing and even the legal field, these systems can enhance a company’s progress in ways nothing else can. As a future lawyer, keeping those technologies in mind, and even familiarising yourself with how to operate them would be a great advantage to have in a world that is moving further away from humans having to perform menial tasks. Instead, high-level critical thinking, decision-making and communication skills will become even more valued in the working world.
Looking to develop your legal skills? Check out our article on 10 Important Lawyer Skills and How To Develop Them.
Loading More Content