Science fiction movies have predicted what we are experiencing today. In the last decade, much has been written on artificial intelligence (AI) and "show robots."
But no one predicted that machine intelligence will cross into the world of drawing, poetry, writing, and journalism through a robot that (in principle) lacks the emotions and feelings needed for this type of expression.
Since November 2022, ChatGPT and the company that introduced it, Open AI, has wreaked havoc on Google. Google, the #1 search engine in the world, has not had a competitor like this in over two decades.
With the value of its activities of $149 billion, Google processes about 100,000 searches per second and provides essential, direct, and accurate answers thanks to its smart algorithm. All other search engines combined represent a tenth of the searches on Google daily.
The $10 billion that Microsoft has invested in OpenAI may have forced Google CEO Sundar Pichai to declare "Code Red" in the company.
Google founders Larry Page and Sergey Brin, who have been away from daily management since 2019, started intensive meetings with the company's senior managers late last year to switch business gears.
Since the news on Microsoft's intention to invest new billions into OpenAI — adding to the $1 billion invested on the day of its founding eight years ago — leaked last December, the company and its renewed old project has garnered a lot of interest from investors, the media, young people, and academics.
Such a massive investment at a time when globally renowned IT companies are suffering unprecedented losses has raised eyebrows. Tens of thousands of their employees were laid off, including giants like Meta, Twitter, Amazon and Google.
Additionally, disaster struck cryptocurrency exchanges and their cardboard companies, especially after the bankruptcy earthquake of the "FTX" company and the exposure of its owner, the "emperor of encryption," the young fraudster Sam Bankman-Fried.
As the news spread, OpenAI was growing at an extraordinary rate. This shows that innovation is still happening, that new chatbots are getting smarter, and that they are preparing to take over the "throne of artificial intelligence" soon.
Google, whose revenues in the last fiscal year amounted to $256 billion, felt threatened by the new smart chatbot and rushed to unveil around 20 new products and show a copy of its search engine with similar and perhaps superior features in the current year.
ChatGPT became the most controversial and problematic tech development since the beginning of this year. It is likely that young people, students particularly, will be eager to use the application which had one million users in only five days of its launch, breaking all previous records for Facebook, Twitter, Netflix, Instagram, and TikTok.
WATCH: ChatGPT, the popular chatbot from OpenAI, is estimated to have reached 100 million monthly active users in January, just two months after launch, making it the fastest-growing consumer application in history, according to a UBS study https://t.co/R65Vpanok9pic.twitter.com/UGxiVON6st
Under pressure from the rise of ChatGPT, Google rushed on 7 February to launch the BARD chatbot that uses the "LaMDA" language model. Still, it got disappointing results, prompting Pichai to order his programmers to allocate about four hours daily to improve the product.
Microsoft also announced an investment of $300 million in another competing programme for ChatGPT. According to the Financial Times, using the AI Claude model, it will acquire a 10 per cent stake in Anthropic, a company founded by former OpenAI researchers.
Unfortunately for Google, a promotional clip for BARD that shared inaccurate information caused a massive sale of shares in Alphabet, Google's parent company, causing a loss of $100 billion in its market value in one session, raising fears that the third largest IT company in the world in terms of market value might lose its position in front of its competitor, Microsoft.
Reuters was the first to pin down an error in Google's announcement of the BARD programme regarding a satellite that took pictures of a planet outside Earth's solar system for the first time.
The technology giant, Google, has been working on similar technology for years, but with greater caution on when and how to announce it, taking into account its possible impact on advertising and e-commerce revenues for the search engine and the risks associated with its use.
The fundamental difference between the Google search engine and ChatGPT is that the former gives hundreds of results for the required search, which may not be necessary for the researcher to find what they are looking for. Chat GPT, on the other hand, provides a more accurate result by formulating a clear and coherent answer, saving the researcher time and effort.
The fundamental difference between the Google search engine and ChatGPT is that the former gives hundreds of results for the required search, which may not be necessary for the researcher to find what they are looking for. Chat GPT, on the other hand, provides a more accurate result by formulating a clear and coherent answer, saving the researcher time and effort.
It looks as if somebody has composed the response and sent it. As the researcher goes on to ask and inquire, they receive answers as if talking to a human being on the other side of the screen.
Recently, Meta CEO Mark Zuckerberg announced that the company has trained a new Large Language Model (LLM), which will soon be released as a service to researchers.
The model LLM is intended to help scientists and engineers explore AI applications such as answering questions and summarising documents.
Self-learning and machine experimentation
ChatGPT is a chatbot that follows a well-known language model developed by OpenAI based on GPT-3.5 (Generative Pre-Trained Transformer), an upgraded version of GPT-3, the most popular AI model for language processing on the Internet, and the latest in the world of Generative AI.
ChatGPT defines itself as being trained on a "big data set including textual data such as dialogues, chats, answering questions, texts from social media, etc.
This allows understanding and responds to human language in a natural way, creating human-generated content. ChatGPT told me, "The data used to train the model includes a variety of texts from different sources and languages (...) so that the model can learn and adapt to new language patterns."
I asked many young users, programmers, and university students about their experience with ChatGPT. Most of the answers ranged between: "Amazing" and "It gives me satisfactory answers" in the year 2021, and some of them underestimated its pre-programmed answers and expected it to fail.
I had a personal experience with "my colleague ChatGPT" and found its answers unsatisfactory from an editorial standpoint, especially when I asked it complex questions. However, it did have an unprecedented ability to understand and respond to my questions and answer me in a few seconds.
I had a personal experience with "my colleague ChatGPT" and found its answers unsatisfactory from an editorial standpoint, especially when I asked it complex questions. However, it did have an unprecedented ability to understand and respond to my questions and answer me in a few seconds.
On a gray and black screen, solutions that sometimes seem surprisingly human are reminiscent of movies such as the 2014 Ex Machina, the 2004 I, Robot, and the 1998 Sphere.
ChatGPT falls under two models. The first is supervised learning, given its mechanism of providing it with data and training it to give answers so that it can then predict the answers itself when injected with new data.
The second is unsupervised learning, which works independently to arrange and classify the data. This stage will be the peak of its independence, which is expected to be reached in the future but needs time.
ChatGPT falls under two models. The first is supervised learning, providing it with data and training it to give answers so that it can then predict the answers itself when presented with new data. The second is unsupervised learning, which works independently to arrange and classify the data. This stage will be the peak of its independence, which is expected to be reached in the future but needs time. .
Engineer and smart cities expert, Muhannad Salam, told Al Majalla that machine learning would take time. The more time we give these machines to learn, most notably, to practice trial and error and correct errors (Machine Learning), the more developed artificial intelligence capabilities will be.
According to Salam, this will not happen overnight — it could take generations just as previous generations did starting with programmes and algorithms and machines that shifted from large room sizes to electronic chips and nanotechnology.
"It is like a young child who needs years to become a doctor, engineer, or philosopher who advises people and writes about them with deep knowledge and success," Salam said.
Companies will start studying ChatGPT successes and failures and will come up with more sophisticated and dependable models.
What is interesting about ChatGPT is that it is trained to use human feedback, a technique called Reinforcement Learning Human Feedback, to understand what people expect when they ask questions.
It uses a combination of chatbots, natural language processing (NLP), machine learning, and deep learning to generate sound, human-produced scripts.
According to Stanford University, GPT-3 contains 175 billion parameters and has been trained on scripts that are 570 gigabytes in size, which is much bigger than its predecessor, GPT-2, which is more than 100 times smaller with only 1.5 billion parameters.
Sam Altman
The ChatGPT project had a bumpy start with its parent company, OpenAI. Founded in San Francisco in late 2015, it is run by Sam Altman, a 37-year-old entrepreneur and former head of startup incubator Y Combinator.
The company has established itself since the beginning by relying on the financial support of prominent shareholders from Silicon Valley, including the co-founder of LinkedIn, Reid Hoffman, investor Peter Thiel, Greg Brockman, Elijah Sotsgiver, John Schulman, and Chuck Zaremba, in addition to Elon Musk, who was on the Board of Directors.
He managed it until 2018 before returning to focus on his Tesla. The latter described the chatbot as "frighteningly good."
Musk has expressed his fears of the negative control of artificial intelligence on several occasions, saying that "Open AI's goal is to democratise artificial intelligence," based on the concept that "freedom stems from the distribution of power, while tyranny results from its limitation."
"It is important, therefore, that the amazing power of AI does not rest in the hands of a few."
Programme already banned
ChatGPT has exceptional potential to help users produce more readable and professional content and text that can be used for light or commercial social media, allowing them to generate huge revenues through advertising and marketing.
However, according to Forbes magazine, its capabilities go far beyond that to include engineering designs, completion of studies, long-term business strategies, marketing plans, and market research by inquiring about the main competitors in any sector.
ChatGPT can pass high-level university exams in various specialties, including medicine, law, and other MBA exams, as stated by the Wharton University in Pennsylvania, which deliberately subjected the chatbot to such a test, as well as creating content and texts for video clips, podcasts, and translation.
Ironically, ChatGPT, which passed the United States Medical Licensing Examination (USMLE), failed miserably in Singapore's Grade 6 exams, achieving 16 per cent in mathematics and 21 per cent in science. However, the chatbot succeeded in the same tests when repeated, proving its ability to develop its own capabilities and learning from its own mistakes.
While some companies plan to integrate the ChatGPT programme into their work, it is being quickly banned in other areas, particularly in the education field: schools and universities.
In New York, for example, the city's education department has ordered the programme to be blocked across all devices and networks in New York public schools. This is due to the real danger of being exploited by students to do their homework and assignments for them.
JPMorgan also barred its employees from using ChatGPT, prompted by concerns about sharing sensitive financial information with the chatbot.
Citi and Goldman Sachs have imposed restrictions around ChatGPT, joining JPMorgan https://t.co/RoI1Wem1mJ
It became the latest entity to restrict the use of chatbots in the workplace after Amazon warned its employees not to share confidential code or information with ChatGPT. Amazon found out that some chatbot responses are similar to internal Amazon data.
Meanwhile, China has banned ChatGBT for not complying with the country's political censorship laws.
On the other hand, Open AI and other companies have introduced anti-fraud programmes that detect fraud in writing research and articles using chatbots, such as GPT-Classifier, AI Writing Check, and GPTZero.
More than 6,000 professors at Harvard, Yale, and Rhode Island universities have signed up for GPTZero, which promises to monitor any text produced by artificial intelligence, according to Edward Tian, the programme's designer and graduate student at Princeton University.
Ethics and intellectual property rights
Although ChatGPT is programmed to refuse to respond to illegal or unethical requests, it may occasionally react to malicious instructions, display discriminatory behaviour, or grant access to false information and conspiracy theories.
To address this, OpenAI has used its Moderation API to warn against or block certain types of content deemed unsafe, such as asking how to bully someone, a violent story, or how to manipulate people or create dangerous weapons.
Intellectual property rights have emerged as one of the most important ethical aspects that must be dealt with — not only in terms of creating content without mentioning its original owners but also for the programme to fabricate research studies and fake sources and associate them with the names of experts or writers who have nothing to do with it, under its mechanism for collecting information.
Intellectual property rights have emerged as one of the most important ethical aspects that must be dealt with — not only in terms of creating content without mentioning its original owners but also for the programme to fabricate research studies.
Assistant Professor of economics at the University of Queensland, David Smerdon tweeted that when he asked ChatGPT about the most cited economic research of all time, he got a false answer by stealing a title identical to an actual book without verifying the real authors.
We may soon witness legal battles between writers and artists on the one hand and artificial intelligence companies over intellectual property rights that reflect the value of human creativity.
Fierce competition awaits
Back to Google, let us recall the story of the expulsion of one of its engineers, Blake Lemoine. Lemoine started making headlines in July 2022 when he said LaMDA showed human-like consciousness.
There are more than 300,000 chatbots in use on Facebook Messenger alone. There are many other examples of similar software developed by Microsoft, Amazon, and Stanford University, all of which have received much less attention than OpenAI or Google.
ChatGPT, albeit relatively recent in artificial intelligence and technology, is distinct because it will pose a challenge to the search engine Google, as Open AI expects to reap $200 million from its new system in 2023 and $1 billion in 2024, according to Reuters.
In another competition for Google, Microsoft has also announced including ChatGPT in its Bing search engine. However, the lousy performance prompted Bing to limit conversations to only five questions at each use and 50 questions per day.
The inaccurate, suspicious, and confusing answers that some users obtained raised doubts about the success of this step.
Microsoft expects that for every additional percentage it will gain from its market share due to the search engine, its annual advertising revenues will grow by $2 billion.
Perhaps it is too early to judge what Artificial Intelligence will carry in the future, as talking about the professional version (ChatGPT Professional) may not be able to cover the repercussions to our lives as individuals and companies.
So far, Open AI itself is still unaware of the size of the interaction and participation that the new version will harvest, as distinguishing between serious users and those enthusiastic and curious viewers is not available yet.
Perhaps this new machine can print, narrate or collect millions of narratives and gain billions, but it will never be able to compose one novel with a human dimension or write a great universal poem.
The words written with ink misgivings, blood, sweat, feelings, and the subconscious will remain more substantial than any machine, no matter how smart its intelligence is.
Most importantly, we should be aware of the implicit risks of these algorithms to the privacy of individuals and societies, and to the intelligence of future generations.
