Anthropic vs China and the question of ‘theft’ with Anthropic’s high-stakes IPO underway

Anthropic has accused China of “distillation” of its Claude AI model using thousands of fraudulent accounts. China denies the fraud. This issue has become a perennial problem, with AI training getting non-stop flak from the owners of source materials used for training. The accusation is the most high-profile yet.

The immediate result was that Alibaba, accused of theft, saw its stocks slide 4.9%. That’s a pretty expensive example of outrage, if it is one, and the question of coincidence is pretty thoroughly unanswered and unanswerable. Remember that all we actually have here are allegations and denials.

Volume of trade during the fall was clearly much higher than average trading. Even the suggestion was enough to trigger a market reaction. This is a whole new dimension of problems emerging at the market level, and it can’t be the last.

Distillation is a long-running and perfectly valid sore point in AI. It typically applies to training smaller AI models, the consumer-range AIs, rather than the much larger AI models, which are better used for high-performance Big Data operations.

The sensitivity over distillation has real and potentially gruesome applications in the marketplace. Training defines the capabilities of AI models. Anthropic recently settled with claimants regarding the use of source materials. Any question of significant liability will get instant market scrutiny.

Anthropic IPO and a new frontier in market credulity

Meanwhile, the big Anthropic IPO is already quietly underway. This latest conflict between AI superpowers could become very messy quickly if and when AI performance and market share are affected by distillation issues. It could also set the stage for a long-drawn-out “siege” of AI training models and methodologies.

There’s a lot at stake here, and it’s not just about investor money. What the market believes is the true hard currency. The market is buying AI very much on a Next Big Thing basis. It’s not as if investors are experts in AI or even know much about it. The market responds to good and bad prompts and usually overreacts.

The Anthropic IPO has key triggers for valuation. Distillation effectively undercuts technological leads. It devalues superior AI models and allows less advanced AI a free ride. That’s Anthropic’s perspective.

From any competitor’s perspective, marketing a demonstrably inferior product is suicidal. It’s also undeniably stupid to allow their own products to devalue. You can assume nobody in the AI sector wants to be Brand X.

This is the market perspective, and it will stick with Anthropic as a market leader. Expectations are extremely high, and when market expectations are not met, prices fall, sometimes dramatically. It’s likely to be a bumpy ride.

Anthropic’s evaluation of expectations is almost too indicative

Meanwhile, the entry of AI into the consumer market is a macro-scale potpourri of perspectives. Anthropic recently did a global survey of “What 81,000 people want from AI,” and the entire spectrum of responses is represented. It’s a virtual census of the knowledge bases of users.

This survey deserves some credit as a snapshot of the reality of the Anthropic entry into the wider market. People are worried, confused, positive, negative and effectively lost in the maelstrom of new experiences.

Let’s not forget that AI is also not yet fully integrated or accepted into the world.

It’s still a novelty.

It’s still clunking about with gimmicky chatbots and agents making headlines that are more annoying than reassuring.

People either need it or don’t need it.

They trust it or don’t trust it.

That’s a tricky environment and a moving target for the world’s first major AI IPO. Fear is not a market asset. AI is simultaneously seen as a threat and as an unavoidable part of the future. Fear of Missing Out has taken on a whole new meaning, too. The message is that you must not miss out, despite your fear.

In the higher echelons of pure functionality, like the Cloud, AI is a major asset. Its applications are clearly valuable. User skills, however, are not on that level. Using AI to read contracts makes sense until it misreads the contracts. Reporting is dependent on external sources of data.

Global AI is here

That level of acceptance doesn’t necessarily translate into the consumer market. Whether the AI is American, Chinese, or European, a lot of common factors will define the global state of AI in the future.

Niche markets: AI will have to specialise and diversify to meet specialist demands. This directly affects training, distillation, and operational capabilities.

The sciences: The most diverse and demanding of all markets for AI, the sciences, will lap up new research and development capabilities. They need to do this, largely due to the excruciating costs of their work.

Global media, the arts, and entertainment: The current incredibly cost-heavy state of global media production will naturally gravitate to AI. Even the highly resistant gaming sector is adopting a human+AI form of AI applications for purely practical reasons. The drive for cost and time efficiencies makes AI a huge asset. It might, ironically, even divert money from production to creative sources.

Retail: Online retail can and does benefit financially from streamlined, almost blockchain-level trade.  At the consumer level, AI may be very much an acquired taste, but it does deliver value in terms of communications with sales points and resolution of issues.

Security: The roaring demand for better security is one of Anthropic’s huge market advantages, and it will be an ongoing, defining issue for global AI. This is a true arms race, and losing isn’t an option.

The net effect of this inevitable spread of AI at all levels of applications is to create a need for common platforms and common ground. Global AI is likely to coalesce into a series of common practices and adapt accordingly.

The current major earthquake-like effects of AI will have to settle into workable forms simply so people can get on with their lives. Anthropic’s IPO is the first test of the real global market.