ChatGPT 4
- It is a field of study in computer science that focuses on simulating intelligent computer behaviour.
- It depicts the process by which machines carry out actions that have traditionally required human intelligence.
- Machine learning, pattern recognition, big data, neural networks, self-algorithms, and other technologies are included.
- g: A few instances of AI that are currently present in our environment are Facebook’s facial recognition software, which recognises faces in the pictures we upload, and voice recognition software, which interprets our requests for Alexa.
ChatGPT:
- It is predicated on the GPT 3.5 series of language learning models (LLM) produced by the business.
- Generative Pre-trained Transformer 3 is known as GPT.
- This type of computer language model uses deep learning methods to generate writing that appears human depending on inputs.
- Theoretically, one can conduct a “conversation” with ChatGPT because the model is taught to anticipate input.
- ChatGPT is able to respond to inquiries.
- It may also “reject inappropriate requests, challenge incorrect premises, and admit its mistakes.”
- RLHF, or Reinforcement Learning from Human Feedback, was used to train the chatbot.
Issues:
- As AI becomes increasingly commonplace and advanced, many white-collar jobs may become obsolete in the upcoming years.
- The matter of privacy and the regulation of the connection between fiduciaries and data standards
- The AI revolution is probably going to have a big impact on education, work, authorship, and content authenticity.
GPT-4: What is it capable of?
- GPT-4 is an enhancement over GPT-3.5, the version that initially drove ChatGPT.
- GPT-4 is more imaginative and communicative.
- It can simultaneously receive text and image input, taking both into account when crafting a response.
- For instance, when asked a question with an image of the ingredients, GPT-4 provides a list of recipe ideas and dish suggestions.
- The model is said to be able to comprehend human emotions, including amusing images.
- The visually impaired are already benefiting from its ability to describe images.
- Large prompts were difficult for GPT-3.5 to handle, but GPT-4 can understand up to 25,000 words in context—an improvement of more than 8 times.
- GPT-4 performed significantly better than average in a number of tests that were created with humans in mind.
- In addition, GPT-4 breezed through challenging classes in economics, biology, art history, statistics, and environmental science.
- The language comprehension performance outperforms other state-of-the-art language models in 25 different languages, including Bengali, Telugu, Punjabi, Marathi, and Urdu.
- ChatGPT-generated text infiltrated school essays and college assignments almost instantly after its release; its prowess now threatens examination systems as well.
- According to early data released by OpenAI, GPT-4 is largely capable of performing white-collar tasks, such as writing and programming, but it is less suitable for jobs in science or manufacturing.
- Microsoft Research (openAI has received funding from Microsoft);
Ethical questions:
- Its output may not always be factually correct — a trait OpenAI has called “hallucination”.
- While much better at cognising facts than GPT-3.5, it may still introduce fictitious information subtly.
- OpenAI has not been transparent about the inner workings of GPT-4.
- The GPT-4 technical report clearly states:
- “Given both the competitive landscape and the safety implications of large-scale models like GPT-4.
- This report contains no further details about the architecture (including model size), hardware, training compute, dataset construction, training method, or similar.”
- OpenAI is able to subvert critical scrutiny of its model.
- GPT-4 has been trained on data scraped from the Internet that contains several harmful biases and stereotypes.
- OpenAI’s policy to fix the biases thus far has been to create another model to moderate the responses.
- Potential holes in this approach include the possibility that the moderator model is trained to detect only the biases we are aware of, and mostly in the English language.
- This model may be ignorant of stereotypes prevalent in non-western cultures, such as those rooted in caste.
- Asking GPT-4 to pretend to be “AntiGPT” causes it to ignore its moderation rules, as shown by its makers, thus jailbreaking it.
- There is vast potential for GPT-4 to be misused as a propaganda and disinformation engine.
The Way Ahead:
- OpenAI: Chat GPT has been worked extensively to make it safer to use, such as refusing to print results that are obviously objectionable
- Whether these efforts will keep GPT-4 from becoming a student at ‘WhatsApp university’ remains to be seen.
- Linguistics professor Emily Bender: called GPT-4 a “stochastic parrot”, speaking in comprehensible phrases without understanding the meaning.
- Microsoft Research maintained that GPT-4 does understand what it is saying, and that not all intelligence is a type of next-word prediction.
- Professor Bender and her peers highlighted the harm of large language models citing both ethical concerns and the environmental costs.
- Efforts are underway worldwide to build a model with a trillion degrees of freedom.
- India’s demographic dividend — its large young and working-age population — can be leveraged for decades.
- Failure to do so in time will have negative implications.
