#036 Friday Special - OpenAI Might Shut Down ChatGPT, AI For Carbon Removal, Smaller LLMs for Generative AI, Anthropic Raises $100M.
Fresh & Hot curated AI happenings in one snack. Never miss a byte 🍔
This snack byte will take approx 6 minutes to consume.
AI BYTE # 1 📢 : It Is Possible That OpenAI Might Shut Down ChatGPT, The Most Popular AI Chatbot?
⭐ ChatGPT is not profitable for OpenAI or its main investor, Microsoft. In fact, it is costing them a lot of money and resources to keep it running.
OpenAI might just have to pull the plug on ChatGPT soon.
The Cost of Running ChatGPT
ChatGPT is not cheap to run. According to The Information, OpenAI had to bear the loss of $540 million to build ChatGPT, and the operational costs are exorbitantly high as well – around $700,000 per day. That’s a lot of money for a free service that is not generating any revenue.
The main source of income for OpenAI is the Azure OpenAI Service, which allows developers to use ChatGPT and other GPT-based products on Microsoft’s cloud platform. But this service is not very popular compared to the free web version of ChatGPT, which has over 1.5 billion users.
Microsoft has already integrated ChatGPT into Bing Search, but it has not been very successful in attracting users. Microsoft might want to push all the users of ChatGPT to its own services, instead of freely giving it away to everyone.
The Data Privacy Issue
Another problem with ChatGPT is the data privacy issue. Many users are concerned that OpenAI might be collecting their personal data from their conversations with the bot.
OpenAI has been trying to address this issue by allowing users to opt out of data collection and by partnering with news and journalism websites to get access to first-hand information. But these measures are not enough to satisfy the users or the regulators.
Many companies have banned the use of ChatGPT for their employees due to privacy concerns. Many users have also decided to migrate and build their own language models using open-source offerings such as Llama 2.
The Ethical and Legal Troubles
OpenAI has been facing a lot of criticism and lawsuits for creating ChatGPT. It has been accused of violating intellectual property rights, spreading misinformation, and enabling cybercrime. It has also been questioned for its transparency and accountability.
OpenAI has been trying to fix these issues by adding filters and warnings to ChatGPT, but they are not very effective or reliable. It has also shut down its AI classifier for identifying AI-generated text because it admitted that it was not able to do the job.
The Future of ChatGPT
All these factors point to the fact that ChatGPT might not survive for long as a free web service. It might be shut down by OpenAI or acquired by Microsoft in the near future.
This does not mean that ChatGPT is a failure or a waste of time. It has been a remarkable achievement in AI research and development.
It has shown the potential and the limitations of language models and conversational agents. It has also inspired many people and sparked many innovations.
AI BYTE # 2 📢: What Happens When You Combine AI With Carbon Removal?
⭐ Artificial intelligence is not just a buzzword in the tech world. It is also a powerful tool for tackling one of the biggest challenges of our time: Climate Change.
According to Crunchbase data, more than $400 million went to dozens of seed- through late-stage rounds straddling these two focus areas in the past two years.
These startups are using AI to track, measure, and reduce carbon emissions across various sectors, such as energy, transportation, agriculture, and manufacturing.
Some of the notable examples are:
Sylvera, a London-based climate software startup that applies machine learning and geospatial data analytics to assess carbon projects. It raised $57 million in a July Series B led by Balderton Capital.
Pachama, a San Francisco-based startup that uses satellite data and AI software to help companies vet and invest in carbon credits. It raised $55 million in a Series B last year, drawing a mix of climate VCs and celebrity backers.
Saildrone, an Alameda, California-based company that operates a fleet of unmanned boats that collect ocean data. It raised $189 million to date, citing oceanic ecosystem monitoring as one of its core use cases.
Arloid Automation, a London-based startup that deploys AI to move buildings closer to net zero. It raised $10 million in a Series A this year, claiming to save up to 40% of energy costs for its customers.
These are just some examples of how AI and carbon reduction are creating new opportunities for innovation and impact.
As the world faces the urgency of addressing climate change, we can expect to see more startups leveraging AI to find solutions that are scalable, efficient, and sustainable.
AI BYTE # 3 📢 : How to Use Smaller Language Models for Generative AI in Business?
⭐ Generative AI is a powerful technology that can create new content from natural language inputs. It can help businesses with tasks like generating marketing campaigns, product ideas, customer insights, and more.
But how can you use generative AI without compromising your security, privacy, and budget?
One solution is to use smaller language models, which are pre-trained on general vocabulary and speech, and then customized with your own data and domain knowledge.
Smaller language models are faster, more accurate, and more cost-efficient than large language models like ChatGPT and Bard, which are trained on billions of parameters from the internet.
Smaller language models can be operated within your secure cloud environment, so you don’t have to worry about exposing your sensitive data to public LLMs.
You can also combine your data with third-party data sources that are relevant to your business goals, such as economic indicators, weather forecasts, or industry trends.
There are several options for smaller language models that you can use for generative AI in business, such as AI21, Reka, Alpaca, and Vicuna.
These models can be easily integrated with your existing data platforms and tools, such as Snowflake, which offers a comprehensive solution for data management and analytics.
AI BYTE # 4📢 : Anthropic Raises $100M From SKT To Build A Custom AI Model For Telcos.
⭐ Anthropic, the AI startup behind the ChatGPT competitor Claude, is on a roll. After raising a whopping $1.5 billion in May and getting support from SAP, the company has secured another $100 million from South Korean telecom giant SKT.
But this is not just another investment. It’s a strategic partnership that will see Anthropic develop a custom large language model (LLM) for the telecom industry, supporting six languages: Korean, English, German, Japanese, Arabic and Spanish.
The LLM will be fine-tuned for different telco-specific use cases, such as customer service, marketing, sales, and interactive consumer applications.
It will also be integrated into the Telco AI Platform, a joint initiative by SKT and its global telco partners: Deutsche Telekom, and Singtel.
This is a huge deal for both Anthropic and SKT. Anthropic will get to showcase its state-of-the-art AI technology, including its constitutional AI approach that aims to create ethical and trustworthy AI systems.
SKT will get to leverage Anthropic’s AI expertise and its own Korean language LLMs to create a competitive edge in the AI ecosystem.
I am keen to see how this collaboration will unfold. Will Anthropic’s LLM outperform the general models on the market?
Will SKT and its partners be able to launch innovative and reliable AI services for their customers? Will this partnership pave the way for more industry-specific LLMs in the future?
I guess we’ll have to wait and see. But one thing is certain: Anthropic is not slowing down anytime soon.
With over $1.6 billion in funding and a vision to transform the world with AI, the company is definitely one to watch.
You can't stop innovation.
Either you become part of it, or you cease to exist