writer.com CEO May Habib attends “Harper’s Bazaar At Work Summit” organized in partnership with Porsche and One&Only One Za’abeel at The OWO London Raffles Hotel on November 21, 2023 in London, England.
Dave Bennett | Getty Images
Writer, a San Francisco-based artificial intelligence startup, launches a massive AI On Wednesday, the model will compete with enterprise products from OpenAI, Anthropic and others. But unlike some competitors, it doesn’t need to spend as much money to train its artificial intelligence.
The company told CNBC it spent about $700,000 to train its latest model, including data and GPUs, while rival startups spent millions building their own models. Its strategy has caught the attention of investors.
Writer will raise up to $200 million at a $1.9 billion valuation, according to a person familiar with the matter who spoke to CNBC. That’s nearly four times the company’s valuation last September, when it raised $100 million at a valuation of more than $500 million.
The company uses synthetic data, or data created by artificial intelligence, to cut costs. It is designed to mimic the real-world information typically fed into models without compromising privacy, and is becoming a more popular training method.
A revised June study by artificial intelligence researchers found that if current AI trends continue, tech companies will “completely exhaust” publicly available training data Between 2026 and 2032wrote that “human-generated public text data cannot sustain beyond this decade.”
Amazon Synthetic data is used when training Alexa, Yuan It has been used to fine-tune the Llama model, MicrosoftSupported OpenAI is reportedly incorporating it into its models job description Published by the company. However, some experts warn that synthetic data should be used with caution as it has the potential to degrade model performance and exacerbate existing biases.
Writer co-founder and chief technology officer Waseem Alshikh told CNBC that Writer has been developing a synthetic data pipeline for several years.
“There is some confusion in the industry about the definition of ‘synthetic’ data,” Alsheikh said. “To be clear, we don’t train our models on fake or hallucinated data, we don’t use models to generate random data… We take real factual data and transform it into specifically structured Synthetic data A clearer, cleaner way to train models.
The company’s generative AI allows enterprise customers to use its large language models (LLMs) to generate human-sounding text for anything from LinkedIn posts and job descriptions to mission statements and data analysis and summaries. The company has more than 250 enterprise customers, including Accenture, Uber, Salesforce, L’Oréal and Vanguard, who use the technology in support, IT, operations, sales and marketing.
The generative artificial intelligence market is poised for growth Up to $1 trillion Income over ten years. Investors have poured $26.8 billion into 498 generative AI deals so far in 2024, and companies in the industry raised $25.9 billion in 2023, a more than 200% increase from 2022, according to PitchBook.
