Meta trained AI on FB and IG posts

Nothing to be surprised at here I don’t think: Meta Platforms used public Facebook and Instagram posts to train parts of its new Meta AI virtual assistant, but excluded private posts shared only with family and friends in an effort to respect consumers’ privacy, the company’s top policy executive told Reuters in an interview. Meta also did not use private chats on its messaging...

Blue Light glasses are probably useless

Summary from this Guardian article: there doesn’t seem to be much benefit from wearing blue light filtering glasses. Or for that matter, reducing your screen time before bed. At least not due to blue light emissions anyway. But reading something stressful, or watching something exciting… that will likely impact your sleep no matter how much blue light is involved. I have a few friends...

Meta AI quietly grows, while Jasper declines

Wow. Meta just keeps rolling out the releases this year. This week Llama 2 expands with Llama 2 Long and rubs shoulders with ChatGPT 3.5 Turbo (not 4.0 to be clear) and Claude 2 in terms of context windows. Meanwhile, Jasper feels the heat and lowers its projections and internal valuation (to $1.2B). This, hot on the heels of OpenAI announcing a staggering internal valuation (of $90B) the other...

OpenAI valuing itself around $90B

Via a report in the Wall Street Journal, OpenAI is looking to allow staff to sell shares at prices that value the company in the $80-90B mark. Note: there’s no intention for the company to go public. To put that in perspective, in Australia, there are only two companies who have a higher market cap. Put another way, OpenAI is valuing itself more than almost every company in Australia...

Amazon invests in Anthropic

Perhaps not surprising that Amazon has invested heavily in Anthropic (makers of Claude).

Meanwhile ChatGPT releases another big update, rolling out voice and image capabilities.

What a year it has been so far, and it still feels like we’re just beginning.

AI and Legal

A few related AI articles. First up, Reuters reporting that a trade group for US authors is suing OpenAI, accusing the company of unlawfully training ChatGPT on authors’ books. The report is catching attention because the authors represented include John Grisham, Michael Connolly and George R. R. Martin: “The complaint said ChatGPT generated accurate summaries of the authors’...

CRMs and Connection

Most CRMs will highlight how they enable you to connect with your prospects and customers better. They help you ‘engage’. This can be good and bad. On the good side, it can keep you organised, so you’re freed up to care about people and provide value. On the bad side, it promotes an attitude of insincere activity – in the name of ‘staying connected’. Example:...

AI Funding

Just noting this here as a timestamp more than anything – AI writing startup Writer raises $100M at a $500M valuation. So many stories like this.

We’re heading into crazy times again. Buckle up.

Assume Everything will be Exposed at some point

You may have seen the TechCrunch article about Microsoft researchers accidentally exposing terabytes of sensitive internal data when providing access to some AI training data. In summary: This data included 38 terabytes of sensitive information, including the personal backups of two Microsoft employees’ personal computers. The data also contained other sensitive personal data, including passwords...

Unity and Communication

I don’t use Unity and don’t really understand what it does, but boy do I know all about the outrage their new pricing model has generated.

A classic example of poor communication causing havoc. The reputation hit is significant.

Another case of perception being more important than reality.