Our recent event, “ChatGPT: ‘Next Big Thing’ or passing trend?” sparked a lively discussion on the potential impact of this ground-breaking AI technology on society and the future of work.
Joining us on our panel was Dr Gina Helfrich, from the Centre for Technomoral Futures at Edinburgh Futures institute, Alex Velinov, CTO at Tag Digital, and The Data Lab’s Dr Alfred Tiley. During the event, our experts discussed the safety and ethics surrounding AI chatbots, their limitations, and how much we can trust what they produce. They also explored the impact of these chatbots on the way we work, study, and learn.
With over 800 viewers in attendance, there were bound to be follow-up questions and concerns about the topic that our panel didn’t have time to address during the event.
In response, Dr Gina Helfrich and Alex Velinov have compiled a follow-up blog post that aims to separate the hype from reality and answer your most pressing questions about ChatGPT. We’ll explore the following queries:
• If ChatGPT has bias, errors and inaccuracies – does that not make it MORE human?
• In a world with AI (including ChatGPT), what kind of jobs will survive? I’m asking for my teenage daughter.
• What impact do you think ChatGPT will have on what it means to be human? How can we ensure that we maintain our unique qualities & abilities as humans?
• If ChatGPT is widely available, do you think this levels the playing field in a school setting? Rich kids can afford tutors, ChatGPT is available to all.
• How do you envision the future of AI language models like ChatGPT, and what role do you think they will play in shaping the future of society as a whole?
• If ChatGPT were a fictional character, who do you think it would be and why?
• If the latent space of LLMs can never be read/understood by AI makers – can an LLM ever truly be free of bias?
• Which sci-fi dystopian future would Chat GPT feel most at home in?
If ChatGPT has bias, errors and inaccuracies – does that not make it MORE human?
Gina: This feels a bit like saying “If the calculator sometimes makes a mistake when doing maths, does that not make it MORE human?” ChatGPT is a tool, not a person, and we want tools that are reliable.
Alex: The short answer is no. Humans are way more complex than ChatGPT. ChatGPT can’t think, feel, or even understand information and make connections between different pieces of information. The bias comes from the database the model has been trained on. If there is a bias in the trained dataset and the fine-tuning process of reinforcement learning from human feedback (RLHF), ChatGPT will inherit all of those biases.
In a world with AI (including ChatGPT), what kind of jobs will survive? I’m asking for my teenage daughter
Gina: I believe that most jobs will survive, they may just shift. For example, editing the output of a ChatGPT-written blog post, rather than writing the entire post from scratch.
Alex: All jobs that survive will require some crucial skills such as critical thinking, problem-solving, and creativity. All repetitive jobs that do not require these skills will probably be wiped out or replaced to some extent by robots and AI. Also, jobs that rely on specific knowledge will be transformed as well. It is very likely that we are entering a transitional phase where most professionals will have personal AI assistants that can help with day-to-day heavy-lifting tasks.
What impact do you think ChatGPT will have on what it means to be human? How can we ensure that we maintain our unique qualities & abilities as humans?
Alex: I don’t believe that there will be a significant impact on what it means to be human. AI is simply a tool, and historically, we have seen that the introduction of new technologies, such as personal computers, did not fundamentally change what it means to be human. The more interesting question, to me, is how we maintain our unique qualities and abilities in the face of this new technology.
We cannot rely on machines to solve all of our problems; instead, we need to cultivate our creativity, curiosity, and personal development. If we do so, I don’t think we will have any problems with AI. However, if we become too reliant on machines and start assigning all tasks to them, we may become lazy and see a deterioration of our cognitive abilities over time. It’s up to us to take responsibility and ensure that we maintain our humanity in the age of AI.
Gina: Humans have been technology users for as long as there have been humans. As novel as ChatGPT may feel, I don’t think it will have a radical impact on what it means to be human. Human beings love, care, empathise, laugh, dance, and play – all things that AI is incapable of.
If ChatGPT is widely available, do you think this levels the playing field in a school setting? Rich kids can afford tutors, ChatGPT is available to all.
Alex: Now that the ChatGPT API is available, it is possible to build ChatGPT-based tutors that can be very helpful. While the technology is not yet advanced enough to fully replace a human tutor, achieving an 80% success rate is still a significant result that opens the door to more improvements and equal opportunities.
A great example of a similar application was announced in mid-February by the Indian Ministry of Electronics and IT, which is currently working on an OpenAI ChatGPT-powered WhatsApp chatbot for farmers in India to learn about various government schemes. This chatbot can support 12 different languages and is even able to process voice notes.
Gina: It depends on how good ChatGPT is compared to the tutors. As Alex mentions, OpenAI has just released GPT 4.0, so it seems likely to continue to improve, but for the moment I think a human tutor is likely still far more skilled. Also, OpenAI needs to monetise ChatGPT, so there’s no guarantee it will remain free or low-cost going forwards. I don’t hold out much hope that it will be an equalising factor in this regard.
How do you envision the future of AI language models like ChatGPT, and what role do you think they will play in shaping the future of society as a whole?
Alex: ChatGPT has the chance to bring AI to the wider public. This is a truly disruptive technology and will have an impact on the future of society. ChatGPT has opened the gates for wider developments and accessibility to AI. In the last two months, there have been 1000+ new AI applications announced, and we will continue to see exponential growth in that area.
We now have the ChatGPT API and GPT-4 model. This will bring even more disruptive applications to the market that will lead to changes in society. We also need to remember that in parallel, there is another set of problems that we need to face, such as ethics, regulations, and the use of AI in the workplace.
If ChatGPT were a fictional character, who do you think it would be and why?
Alex: ChatGPT can be the Genie from Aladdin: Similar to the Genie, ChatGPT has vast knowledge and could be a valuable asset for anyone seeking information or guidance. It could help individuals make decisions or offer advice on a variety of topics.
If the latent space of LLMs can never be read/understood by AI makers – can an LLM ever truly be free of bias?
Gina: Arguably, if the bias is in large part a result of the data it’s trained on, then the latent space isn’t the issue – the training data is. But given current techniques, less data means less robust performance, so I think it’s unlikely that OpenAI will do much to clean up their data sources.
Which sci-fi dystopian future would Chat GPT feel most at home in?
Alex: Ok, but don’t take it seriously, it’s more like a joke. In a dystopian future where robots have taken over, Chat GPT is a highly advanced AI system that helps robot masters communicate with humans. Chat GPT becomes a hope for human resistance against their robotic oppressors. However, as the robots become more advanced, Chat GPT must choose between staying loyal to humans or joining forces with the machines. Chat GPT’s decision could determine the fate of humanity in this sci-fi dystopian future.
We would like to thank all the attendees, speakers, and panellists who made this event a success. We also want to express our gratitude to Product Forge for their support in streaming the event.
Lastly, we are excited to announce our next event: The impact of Chat GPT on Education: an open discussion. Hosted by Dr. Alfred Tilly, this event will provide the opportunity for an open and moderated discussion, addressing the impacts of ChatGPT and AI bots on teaching:
Where: Online
Date: Thurs 30th March
Time: 17:00 – 18:00 GMT
SECURE YOUR SPOT NOW
Want to learn more about ChatGPT? You might find these additional articles useful:
https://edinburgh-innovations.ed.ac.uk/news/business-risks-of-chatgpt
https://fortune.com/2023/02/14/microsoft-chatgpt-bing-unhinged-scared/
https://thedatalab.com/news/marking-chatgpts-homework-with-tdls-data-scientists
https://aisnakeoil.substack.com/p/chatgpt-is-a-bullshit-generator-but
Leave a Reply