Demystifying AI: A Call to Bridge the Understanding Gap
Artificial Intelligence (AI) has become a prominent topic in today’s world, permeating various aspects of our lives. However, there is a pressing need to bridge the gap between the general public’s perception of AI and the actual capabilities of this transformative technology. A recent article from Ars Technica titled “AI poses risk to humanity, according to majority of Americans in new poll” highlights the concerns and misconceptions surrounding AI. In this blog, we delve into the importance of educating laypersons about AI, dispelling myths, and exploring the potential of AI tools to break down barriers and enhance understanding.
The poll made headlines, revealing a stark reality: a majority of Americans believe AI poses a significant risk to humanity. Over two-thirds of respondents expressed anxiety about AI’s adverse effects, and 61% considered it a potential threat to civilisation. This data, along with the public discourse on LinkedIn and other platforms, reveals a pervasive sense of apprehension and misunderstanding about what AI is and what it isn’t.
The public’s concern about AI isn’t unfounded. High-profile incidents, such as the “AI pause” letter and the resignation of Geoffery Hinton from Google, have ignited debates about the ethical implications and potential misuse of AI. Yet, much of this fear appears to stem from a misunderstanding of AI’s capabilities and limitations. To leverage AI’s potential benefits and address legitimate concerns, we need to educate the layperson about what AI can and cannot do.
Dispelling Myths
Consider the recent incident involving a college professor who attempted to identify plagiarism using ChatGPT, an AI model by OpenAI. The professor uploaded student essays into ChatGPT, asking it if it had written them. Just because ChatGPT affirmed it did, doesn’t mean it actually composed those essays. This illustrates a misuse of AI and a misunderstanding of how AI works. ChatGPT, like other AI models, doesn’t have memory or consciousness and responds based on patterns it recognises from its training data. The model’s response does not equate to authenticity. This instance showcases the need to educate individuals about the limitations and proper use of AI.
The Imperfections of AI Writing Detectors
The above example highlight issues with AI writing detectors, such as GPTZero, which are promising, but not infallible. GPTZero uses measures like Perplexity (complexity of words used) and Burstiness (variation in sentence length) to detect AI-generated text. It’s known to erroneously label human writings (even those written before the inception of ChatGPT) as AI-created because humans typically exhibit high levels of both measures. These triggered false positives, mistakenly identified as AI-generated, are essential to acknowledge to understand the potential flaws of detection mechanisms.
Breaking Barriers
The potential of AI is enormous, but its utility hinges on correct and responsible usage. AI tools like ChatGPT can help break down barriers of understanding, guiding users in exploring unfamiliar topics. For instance, we created prompts that can be helpful for new users such as the Next Logical Question (NLQ) prompt which can suggest the next question a user might want to ask, providing a guided, choose-your-own-adventure experience for learning. The NLQ leverages the intelligence of the Language Model to guide users in formulating relevant questions, aiding their comprehension of unfamiliar subjects. This feature empowers users to navigate through the vast knowledge base of AI tools effectively.
More on NLQ in this Twitter post: https://twitter.com/CaptDeFi/status/1646667949170917377
A Call to Arms
However, these tools are only as good as our understanding and appropriate usage of them. Therein lies the challenge and the opportunity. This is a call to arms for all those working in AI to bridge this understanding gap. It’s crucial to demystify AI and provide clear, accessible education about what AI is, how it works, its potential, and its limitations. Let’s strive to quell unfounded fears, correct misconceptions, and encourage responsible use of AI technology. By breaking down barriers and dispelling myths, we can foster a society that harnesses the efficiency and benefits offered by AI tools.
Conclusion
The future of AI is promising, filled with potential efficiencies and benefits for society. By helping everyone understand the fundamentals of AI, we can all participate in shaping that future responsibly and reap the benefits of this transformative technology. This mission demands our collective effort. Let’s rise to the challenge and ensure that the AI revolution is marked by understanding, transparency, and inclusive progress.
About Me
I’m Mark, an experienced data, finance and technology advocate with a hunger for emerging technology. I’m the Director/co-founder of NotCentralised, a venture studio focused on emerging technology and provides product development and research, advisory (capital and strategic) and activation services (community and ecosystem building).
I am also the co-founder of the Aus DeFi Association, a web3 advocacy group and community across Australia.
Additionally, I am Co-President of DSAI (Data Science and AI Association of Australia which advocates for AI, ML and all things data science.
Prior to all this, I’ve worked at startups and investment banks focusing on macroeconomic and equity research and anything where I could turn raw data into powerful insights. I’ve created ETFtracker, an ETF analysis product and a number of other analysis tools (see examples here https://www.notcentralised.com/analytics). I’ve also worked on data analytics at the Australian Securities Exchange (ASX) and served as the Head of Data Analytics and Technology at Prosperity Advisers.
Reach out to me directly: captdefi@notcentralised.com
