Affiliate links on Android Authority may earn us a commission. Learn more.
I used AI to settle the Android vs iPhone debate, here's how it went
That machines will steal our jobs has been a common fear stemming back to the industrial revolution. We’ve long assumed that to be an inevitable truth for repetitive, manual, labor-intensive jobs. However, as writers or creative individuals, we tend to think that our professions sit well protected in a relatively safe domain. So when DALL-E 2 and ChatGPT were released this year, the potential repercussions of AI hit a bit closer to home.
DALL-E 2 and ChatGPT have brought up fears of computers replacing humans, but the reality is a bit more nuanced.
While DALL-E’s AI-initiated art proved to be an exciting experiment in derived creativity, ChatGPT takes it further. The AI and machine learning-based language model can script human-like content while pulling from a database of information that vastly supersedes human capabilities. Welp. Guess the computers finally came for my job?
To put it to the test, I challenged the algorithm to answer a question that has created and toppled smartphone empires and instigated numerous wars amongst raging fanboys. Hey, Chat GPT, let’s settle the Android vs. iPhone debate once and for all.
It takes the right prompts
As technology journalists covering bleeding-edge tech and writing about how it intersects with our lives, our jobs require context, nuance, and experience. We thrive on our ability to cut through the hyperbole and bring you the low-down. Could an AI, clever as it is, master this craft using the power of practically unlimited computing potential?
ChatGPT's conversational skills are highly dependent on user input and prompts.
Logging into ChatGPT, I had some understanding of what to expect from the algorithmic approach toward text generation. Like DALL-E 2, ChatGPT’s conversational skills are relatively prompt-driven, and it is possible to get a more detailed answer if you frame the question with specifics in mind. I played around with multiple iterations of the question, including asking the AI to help me pick between the best flagship Android phone and an iPhone, assuming that the price was the same. This is what I got:
It’s difficult to say which is better, as both iPhone and Android devices have their own unique strengths and weaknesses. Ultimately, the best choice for you will depend on your personal preferences and needs. Some people prefer iPhones because they are easy to use and offer a seamless user experience, while others prefer Android devices because they tend to be more customizable and offer a wider range of options at different price points. It really comes down to what you value most in a phone.
In more ways than one, the answer presented by ChatGPT was correct. After all, picking out a new phone does come down to what you value most in a phone. The succinct summary was essentially the cliff notes version of what I’d mention informally in a casual chat with a friend looking to buy a new phone. The text is reasonably well presented in a human-like form, even if lacking a bit of emotion.
However, a closer glance reveals the lack of complexity required by a question such as this. The summary offered was perhaps the most basic version of an argument that we at Android Authority and dozens of other publications have written about ad nauseam. What it didn’t give us was any additional intuition or the needed human touch that lets us help pick out the best Android phones at every price point for our readers.
A brilliant search engine by any other name
When asked to elaborate further, ChatGPT didn’t stray too far from the sub-100-word summary above and presented a litany of facts, such as Android phones being powered by the Android operating system and that iPhones are known for their sleek, modern design. That is factually correct, but not very helpful when trying to pick up a new phone.
ChatGPT is great at taking facts and framing them in natural speech, but offers no opinion whatsoever.
In fact, the more you converse with ChatGPT, its limitations become clearer. The AI copy is excellent at regurgitating facts — or the most accepted/prevalent version when there’s nuance — and without a doubt, there are use cases for it. However, I’d place it close to a great example of natural language processing than actual intelligence.
Despite constant prompts, at no point did ChatGPT try to understand what I was looking for in a phone and which device might suit my requirements. As a perfectly neutral referee, ChatGPT excels at stating the rules. But creative expression stems from an opinion, and without the ability to formulate that, it is evident that the tool isn’t going to be much more than a tool in a writer’s arsenal — so far.
You can teach a computer to speak, you can't teach it to think — yet.
ChatGPT’s excellence in forming deceptively good prose isn’t entirely surprising. As a lifelong language learner and a computer engineer by education, it is easy to draw parallels between grammatical structures and coding syntax. Finding the right words isn’t complicated since the algorithm pulls data from an extensive database. Add in thousands of hours of machine learning, and you can teach a computer to speak. What you clearly can’t teach a computer is to have an opinion.
I asked ChatGPT to elaborate further on the topic; even the most in-depth answers were reiterated facts from the thousands of articles on which the AI was trained. No matter the prompt, there was no situation where the AI could nudge me in the right direction, leaving all the actual decision-making to me. This lack of deep understanding of the why or what is perhaps the last remaining frontier of demarkation between human speech and AI-generated text. That said, it is the crucial difference that also separates ChatGPT from being a truly intelligent system versus what it is — a brilliant and intuitive search engine.
It won’t replace humans yet, but it has its uses
For all its limitations, ChatGPT isn’t without its use cases. I already see merit in using the AI-based language learning model to bounce off ideas or pull up contextual information while writing about a topic. Since ChatGPT leans towards highly ranked answers from the broader information it has been trained on, the likelihood of finding the correct information tends to be higher than traditional search engines.
Moreover, it can be a helpful tool to pull up quotes, studies, statistics, or even to find out popular questions that people might be asking about a topic. Additionally, I can see writers using it to craft the outline of an article or find inspiration.
For all its misses, ChatGPT has excellent potential as a tool for creative writers, journalists, and in newsrooms.
Spending a few days playing around with ChatGPT makes it clear: a human replacement it isn’t yet. And factoring in the lack of perspective and opinion ensured that the age-old debate once again remained unsolved. Android vs. iPhone, not even AI has the correct answer.