That's a good question, and one that many people not versed in pensions have. Here's a summary:

Defined benefit, assuming you start from nothing, and greatly simplified

1) The company, through negotiations with the union or employees, defines what the retirement benefits will be.
2) Actuaries are hired to determine what level of funding is required each year in order to ensure that there will be adequate funds available to pay out the future obligations agreed to in #2 (Many variables are in this formula including expected rate of return, average lifespan, etc...)
3) The company makes the required contribution.
4) Funds are managed by an outside financial management firm, based on an agreed upon set of rules/risk profile (also subject to regulation).
5) One year later the company is required to make its next contribution. Additionally, if the funds invested earned less than the expected rate of return used in #2, the company must also make up the short-fall to bring the fund balance to what it should be at year 2.

Defined contribution, assuming you start from nothing, and greatly simplified

1) The company, through negotiations with the union or employees, defines how much they will contribute to each employee's account.
2) The company makes the required contribution.
3) Funds are managed by an outside financial management firm, at the direction of each individual employee (generally they offer a selection of mutual funds to choose from).

Difference

As you can see, in a DB plan, the company bares the risk as they must ensure that funds are available to fund the agreed upon future benefits. In a DC plan, the employee bares the risk as the company is off the hook once it makes the contribution, if the employee doesn't have enough to retire that is their issue.

Over the past few years there have been problems in many DB plans.

1) The Government regulates the funding of plan (i.e. says you must actually have at least XX% of what the plan should have in the account in it). Say that it is 90%. Pretend at the beginning of 2008, you were 100% funded. At the end of 2008, after the crash, you are only 75% funded due to the losses incurred. As a company, you now need to make up AT LEAST 15% of the fund value immediately. Now imagine its a 5 billion dollar plan. That is a very big cheque to write and many companies (including AC) couldn't write it. Such a slide in the market has the ability to bankrupt a company nearly overnight.

2) Future negotiations. Say you've agreed to pay a DB of 70% of each employee's best 5 years starting at age 65. You head to negotiations and settle on changing that to 75% of each employee's best 4 years starting at age 60. Changing each of those number will impact not only what the fund balance needs to be in the future, but what it needs to be TODAY. So even during the 'good old days', an upward adjustment in retirement benefits could result in massive cash outlays right away to get the fund balance to where it needs to be.

Not surprisingly, companies are moving en masse towards DC plans. Non-union environments are pretty much entirely there (Transcanada Pipelines is the only private company that comes to mind that still has DB for new hires). Unions have more power, so the switch at unionized companies has been slower, but is still occurring. The financial impact of this movement won't be seen for decades, but early research is showing that the biggest problem isn't the switch itself, but that companies are changing how much the contribute at the same time. For instance, they might have contributed on average 12% of each employees salary to the pension plan under the DB program, but when they go DC they only contribute 6% of each employee's salary. So even if the employee invested perfectly, it would be impossible for them to match what they would get in a DB plan without investing significant portions of their own money as well.

I could go on for ages about all the pros and cons, but there is a summary.